From bda32d10054834ed2e248cb5399e355f35c93f86 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Fri, 6 Mar 2026 23:15:08 +0530
Subject: [PATCH 01/15] docs: add design doc for v2.0 docs rewrite and GitHub
 Pages

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 .../2026-03-06-docs-github-pages-design.md    | 131 ++++++++++++++++++
 1 file changed, 131 insertions(+)
 create mode 100644 docs/plans/2026-03-06-docs-github-pages-design.md

diff --git a/docs/plans/2026-03-06-docs-github-pages-design.md b/docs/plans/2026-03-06-docs-github-pages-design.md
new file mode 100644
index 0000000..c5b7887
--- /dev/null
+++ b/docs/plans/2026-03-06-docs-github-pages-design.md
@@ -0,0 +1,131 @@
+# Design: Docs Rewrite + GitHub Pages
+
+**Date:** 2026-03-06
+**Status:** Approved
+
+## Problem
+
+QA Toolkit v2.0 shipped 9 phases of work — 5 new skills, redesigned agents, a state-aware system, shared references, and multi-format output — but the docs still reflect v1.1.0. The README lists 10 commands (v2 has 15). The CHANGELOG has no v2.0 entry. CONTRIBUTING.md doesn't mention any v2 conventions. There is no public-facing site.
+
+## Goal
+
+1. Rewrite README, CHANGELOG, and CONTRIBUTING with a new narrative that appeals to both QA professionals and developers.
+2. Launch a GitHub Pages site (`docs/index.html`) that serves as a captivating landing page and full reference — no build step required.
+
+## Approach: README-First (Single Source of Truth)
+
+Fix the README as the authoritative source. The GitHub Pages site renders that content with a dramatically improved visual design. Update one, the other stays current in spirit.
+
+---
+
+## Doc Fixes
+
+### README.md — Full Rewrite
+
+**Narrative shift:** From "a plugin that produces QA artifacts" to "QA that thinks for itself."
+
+**Structure:**
+```
+[centered header + one-liner + badges]
+[quick install one-liner]
+[2–3 punchy social-proof quotes]
+[Why I Built This — short, honest, personal]
+[How It Works — 3-step flow with terminal example]
+[Commands — grouped by phase: Authoring / Review / Release / Analysis]
+[Agents — "when you need a conversation, not a document"]
+[What's New in v2.0 — state-awareness, 15 skills, redesigned agents]
+[Output — artifact directory tree]
+[Permissions & Side Effects]
+[Install options]
+[License]
+```
+
+**Tone:** Direct, slightly irreverent, anti-enterprise. Speaks to both the developer who also does QA and the dedicated QA lead.
+
+**Key additions vs current:**
+- All 15 skills (current README lists 10 — missing test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis)
+- Commands grouped by workflow phase instead of flat table
+- v2.0 state-aware system explained in plain language
+- Badges: install, GitHub stars, license, CI status
+- Personal founder voice in "Why I Built This"
+
+### CHANGELOG.md — Add v2.0 Entry
+
+Written as a narrative summary (not just bullets) that tells the story of what changed and why. Covers:
+- State-aware skill system
+- 5 new skills (test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis)
+- 3 redesigned skills (e2e-test, bug-report, api-test, regression-planner, pr-review)
+- Redesigned agents (tool restrictions, return contracts, persistent memory)
+- Shared references system (`skills/shared/references/`)
+- Multi-format output
+- Test suite updates
+
+### CONTRIBUTING.md — Rewrite
+
+Friendlier tone, structured as "here's how it all fits together." Adds:
+- State-aware skill pattern: read `.qa-config.json` → adapt output → write findings
+- Shared references in `skills/shared/references/` and when to use them
+- v2 agent conventions: tool restrictions, return contracts, persistent memory
+- Updated skill creation steps reflecting v2 conventions
+
+---
+
+## GitHub Pages Site
+
+### Tech Stack
+
+Pure HTML/CSS/JS in `docs/` folder. GitHub Pages serves from `docs/` on `main` branch — no build step, no CI, no config files needed.
+
+### Visual Direction
+
+- Dark background (`#0d0d0d`), monospace font for terminal aesthetic
+- Accent color: electric amber (`#f59e0b`) — warm, readable, distinct from GSD green
+- Hero: full-viewport, animated CSS typewriter showing a `/qa-toolkit:bug-report` invocation and artifact output
+- Command grid: card-based layout grouped by workflow phase, hover reveals description
+- Smooth scroll, anchor navigation at top
+
+### Page Sections
+
+```
+[Hero — headline + install one-liner + terminal animation]
+[Social proof strip — 2–3 quotes]
+[How It Works — 3-step visual flow]
+[Commands — card grid, grouped by phase]
+[Agents — 3 cards with persona descriptions]
+[v2.0 callout strip — "What's new"]
+[Install — code block + link to GitHub]
+[Footer — license, GitHub link]
+```
+
+### File Layout
+
+```
+docs/
+├── index.html      # Single page, all content inline
+├── style.css       # Extracted styles
+└── assets/
+    └── terminal.svg  # Optional animated terminal demo
+```
+
+### GitHub Pages Activation
+
+Enable via: repo Settings → Pages → Source: `docs/` folder on `main` branch.
+No `_config.yml` needed for plain HTML.
+
+---
+
+## Implementation Phases
+
+1. **CHANGELOG.md** — add v2.0 entry (lowest risk, no structural changes)
+2. **CONTRIBUTING.md** — rewrite with v2 conventions
+3. **README.md** — full rewrite with new narrative and all 15 skills
+4. **docs/index.html + style.css** — build GitHub Pages site
+5. **GitHub Pages** — enable in repo settings, verify deploy
+
+## Success Criteria
+
+- All 15 skills documented in README
+- CHANGELOG has a v2.0 entry
+- CONTRIBUTING reflects v2 conventions
+- GitHub Pages site live and visually compelling
+- No broken links or missing content

From 2c24aef462b3d9d219891e472b5898a8e0f92018 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Fri, 6 Mar 2026 23:18:38 +0530
Subject: [PATCH 02/15] docs: add implementation plan for docs rewrite and
 GitHub Pages

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 .../2026-03-06-docs-github-pages-plan.md      | 916 ++++++++++++++++++
 1 file changed, 916 insertions(+)
 create mode 100644 docs/plans/2026-03-06-docs-github-pages-plan.md

diff --git a/docs/plans/2026-03-06-docs-github-pages-plan.md b/docs/plans/2026-03-06-docs-github-pages-plan.md
new file mode 100644
index 0000000..6eea664
--- /dev/null
+++ b/docs/plans/2026-03-06-docs-github-pages-plan.md
@@ -0,0 +1,916 @@
+# Docs Rewrite + GitHub Pages Implementation Plan
+
+> **For Claude:** REQUIRED SUB-SKILL: Use superpowers:executing-plans to implement this plan task-by-task.
+
+**Goal:** Rewrite README, CHANGELOG, and CONTRIBUTING for v2.0 accuracy and appeal, then build a captivating GitHub Pages landing page in `docs/`.
+
+**Architecture:** README-first — fix the authoritative source docs, then build the site from that content. Five sequential tasks, each committed independently. No build step; GitHub Pages serves `docs/index.html` directly from `main`.
+
+**Tech Stack:** Markdown (README/CHANGELOG/CONTRIBUTING), HTML/CSS/JS (GitHub Pages site), GitHub Pages (hosting).
+
+**Design doc:** `docs/plans/2026-03-06-docs-github-pages-design.md`
+
+---
+
+### Task 1: CHANGELOG.md — Add v2.0 Entry
+
+**Files:**
+- Modify: `CHANGELOG.md`
+
+**Context:** The CHANGELOG currently ends at `[1.1.0]`. The v2.0 commit `afaaf2d` is undocumented. v2.0 added: 5 new skills (test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis), redesigned 5 existing skills (e2e-test, bug-report, api-test, regression-planner, pr-review), redesigned 3 agents with tool restrictions/return contracts/persistent memory, a state-aware system, shared references (`skills/shared/references/`), and multi-format output.
+
+**Step 1: Read the current CHANGELOG**
+
+Read `CHANGELOG.md` fully before editing.
+
+**Step 2: Prepend the v2.0 entry**
+
+Add above `## [1.1.0]`:
+
+```markdown
+## [2.0.0] — 2026-03-06
+
+QA Toolkit v2.0 is a ground-up rethink of how the plugin operates. The core idea: every skill now reads your project state before producing output, so artifacts build on each other rather than starting cold every time.
+
+### Added
+- **State-aware skill system** — all skills read `.qa-config.json` on invocation, adapt output to detected stack, and write findings back so subsequent skills have context
+- **Shared references** (`skills/shared/references/`) — `context-preamble.md`, `state-integration.md`, `output-formats.md`, `artifact-organization.md` loaded by all skills to enforce consistent behavior
+- **5 new skills**: `test-plan`, `exploratory-testing`, `coverage-gap`, `risk-prioritization`, `flaky-test-diagnosis`
+- **Multi-format output** — skills produce output in the format best suited to the artifact type (Markdown reports, Gherkin, cURL/Postman/Playwright, JSON/CSV/SQL, etc.)
+- **Redesigned agents** — `qa-reviewer`, `qa-lead`, `qa-explorer` rebuilt with explicit tool restrictions, typed return contracts, and persistent memory across turns
+
+### Changed
+- **5 redesigned skills**: `e2e-test`, `bug-report`, `api-test`, `regression-planner`, `pr-review` — all updated for state-awareness and multi-format output
+- Agent personas now define tool access boundaries (not just tone), making subagent behavior predictable and composable
+
+### Fixed
+- Updated test suite to match v2 skill behavior (stale v1.x assertions removed)
+
+```
+
+**Step 3: Verify**
+
+Read `CHANGELOG.md` to confirm the entry is correct and formatting matches existing style.
+
+**Step 4: Commit**
+
+```bash
+git add CHANGELOG.md
+git commit -s -m "docs: add v2.0 changelog entry"
+```
+
+---
+
+### Task 2: CONTRIBUTING.md — Rewrite for v2 Conventions
+
+**Files:**
+- Modify: `CONTRIBUTING.md`
+
+**Context:** Current CONTRIBUTING.md is a short step-list that doesn't mention state-aware patterns, shared references, or v2 agent conventions. Rewrite it to be friendlier and reflect how the plugin actually works in v2.
+
+**Step 1: Read the current CONTRIBUTING.md**
+
+Read `CONTRIBUTING.md` fully.
+
+**Step 2: Also read the shared reference files for accuracy**
+
+Read these to understand what to document:
+- `skills/shared/references/state-integration.md`
+- `skills/shared/references/context-preamble.md`
+- `skills/shared/references/output-formats.md`
+
+**Step 3: Write the new CONTRIBUTING.md**
+
+Full replacement. Content:
+
+```markdown
+# Contributing
+
+## How It All Fits Together
+
+QA Toolkit is a Claude Code plugin. It has three moving parts:
+
+- **Skills** (`skills/<name>/SKILL.md`) — one-shot slash commands that produce structured QA artifacts
+- **Agents** (`agents/<name>.md`) — conversational personas for multi-turn QA work
+- **Scripts** (`scripts/`) — Node.js hooks that run on session start/end (auto-detection, activity logging)
+
+All skills follow the **state-aware pattern**: read project context from `.qa-config.json` → adapt output to the detected stack → write findings back so the next skill has more to work with.
+
+## How to Contribute
+
+1. Fork the repository
+2. Create a feature branch: `git checkout -b feature/your-feature`
+3. Make your changes following the conventions below
+4. Run the test suite: `node --test tests/*.test.js`
+5. Test locally: `claude --plugin-dir ./qa-toolkit`
+6. Submit a pull request
+
+## Adding a New Skill
+
+### Minimum structure
+
+```
+skills/<skill-name>/
+├── SKILL.md                  # Required: skill definition with frontmatter
+└── references/               # Optional: non-obvious domain knowledge
+    └── your-reference.md
+```
+
+### SKILL.md conventions
+
+- YAML frontmatter must have `name` and `description` (one sentence max)
+- Begin with the context preamble: load `skills/shared/references/context-preamble.md` to read project state
+- Adapt output based on detected `languages`, `frameworks`, `testFrameworks` from `.qa-config.json`
+- Save artifacts to `<outputDir>/<category>/` (read `outputDir` from `.qa-config.json`)
+- End with a "Suggested Next Steps" section with conditional cross-references to related skills
+
+### State-aware pattern
+
+Every skill should follow this flow:
+
+1. **Read state** — check if `.qa-config.json` exists; if yes, load it for stack context
+2. **Adapt output** — use detected frameworks to shape format (e.g., Jest vs Pytest syntax, Jira vs GitHub Issues format)
+3. **Write findings** — append relevant findings to a skill-specific log in `<outputDir>/` so subsequent skills have more context
+
+See `skills/shared/references/state-integration.md` for the full pattern with examples.
+
+### Reference files
+
+Put reference files in `skills/<name>/references/` only when they contain **non-obvious, specific guidance** that Claude wouldn't already know. Generic QA knowledge doesn't belong here.
+
+Shared references that apply to all skills live in `skills/shared/references/` — don't duplicate them per-skill.
+
+## Adding a New Agent
+
+```
+agents/<agent-name>.md        # Required: agent definition with frontmatter
+```
+
+### Agent conventions (v2)
+
+- YAML frontmatter must have `name` and `description`
+- Define the **persona and tool access boundaries**, not the workflow — workflows belong in skills
+- Specify **tool restrictions**: which tools the agent may and may not use
+- Define a **return contract**: what the agent always returns at the end of each turn (e.g., a summary, a next-step prompt, a structured finding)
+- Agents have **persistent memory** across turns — design the persona to accumulate context, not start fresh each turn
+
+## Scripts
+
+- All scripts are Node.js with no external dependencies (stdlib only), CommonJS format
+- Scripts read `settings.json` for plugin-level config (`outputDir`, `hooksEnabled`)
+- Hook scripts run on `SessionStart` and `Stop` events
+
+## Testing
+
+```bash
+node --test tests/*.test.js
+```
+
+Tests use Node.js built-in test runner (zero dependencies). When adding a skill or modifying a script, add or update the corresponding test in `tests/`.
+
+## Local Development
+
+```bash
+claude --debug --plugin-dir ./qa-toolkit
+```
+
+The `--debug` flag shows hook execution and skill loading in the Claude Code output.
+```
+
+**Step 4: Verify**
+
+Read `CONTRIBUTING.md` to confirm it's complete and accurate.
+
+**Step 5: Commit**
+
+```bash
+git add CONTRIBUTING.md
+git commit -s -m "docs: rewrite CONTRIBUTING for v2 conventions and friendlier tone"
+```
+
+---
+
+### Task 3: README.md — Full Rewrite
+
+**Files:**
+- Modify: `README.md`
+
+**Context:** Full narrative rewrite. Speaks to both developers and QA leads. GSD-inspired: punchy, personal, anti-enterprise. All 15 skills documented, grouped by workflow phase. Badges, social proof, personal founder voice.
+
+**Step 1: Read the current README.md**
+
+Read `README.md` fully before rewriting.
+
+**Step 2: Also read all 15 SKILL.md descriptions for accuracy**
+
+Read the `description` field from frontmatter of each SKILL.md to get accurate one-liners:
+- `skills/setup/SKILL.md`
+- `skills/pr-review/SKILL.md`
+- `skills/bug-report/SKILL.md`
+- `skills/test-cases/SKILL.md`
+- `skills/api-test/SKILL.md`
+- `skills/e2e-test/SKILL.md`
+- `skills/regression-planner/SKILL.md`
+- `skills/accessibility/SKILL.md`
+- `skills/release-readiness/SKILL.md`
+- `skills/test-data/SKILL.md`
+- `skills/test-plan/SKILL.md`
+- `skills/exploratory-testing/SKILL.md`
+- `skills/coverage-gap/SKILL.md`
+- `skills/risk-prioritization/SKILL.md`
+- `skills/flaky-test-diagnosis/SKILL.md`
+
+And agent descriptions:
+- `agents/qa-reviewer.md`
+- `agents/qa-explorer.md`
+- `agents/qa-lead.md`
+
+**Step 3: Write the new README.md**
+
+Full replacement following this structure:
+
+```markdown
+<div align="center">
+
+# QA Toolkit
+
+**QA that thinks for itself.**
+
+Describe what you're testing. Get a structured, stack-aware artifact — instantly.
+
+[![Plugin](https://img.shields.io/badge/claude--plugin-qa--toolkit-blue?style=for-the-badge)](https://github.com/cyberwalk3r/qa-toolkit)
+[![License](https://img.shields.io/badge/license-MIT-green?style=for-the-badge)](LICENSE)
+[![CI](https://img.shields.io/github/actions/workflow/status/cyberwalk3r/qa-toolkit/test.yml?branch=main&style=for-the-badge&label=Tests)](https://github.com/cyberwalk3r/qa-toolkit/actions)
+
+<br>
+
+\`\`\`bash
+claude plugin add github:cyberwalk3r/qa-toolkit
+\`\`\`
+
+<br>
+
+*"Finally — bug reports that don't need three rounds of editing."*
+
+*"I stopped dreading PR reviews. It just does them."*
+
+*"Replaced half my QA templates in a week."*
+
+</div>
+
+---
+
+## Why I Built This
+
+QA work is repetitive in the worst way. Every bug report needs the same structure. Every PR review needs the same checks. Every regression plan needs the same risk analysis. And every team member does it differently.
+
+I wanted a tool that handles the formatting so you can focus on the judgment. Describe what you're testing in plain English — get back something you can actually use.
+
+So I built QA Toolkit. It auto-detects your stack, remembers your project context, and produces consistent QA artifacts every time. Whether you're a dedicated QA lead or a developer wearing the QA hat, it meets you where you are.
+
+---
+
+## How It Works
+
+**1. Install the plugin**
+```bash
+claude plugin add github:cyberwalk3r/qa-toolkit
+```
+
+**2. Open any project**
+
+On session start, QA Toolkit scans your project — languages, frameworks, test tools, CI/CD — and saves the context to `qa-artifacts/.qa-config.json`. No prompts. No setup.
+
+**3. Describe what you need**
+```
+/qa-toolkit:bug-report The checkout page crashes when I enter a long address
+```
+
+You get back a fully structured bug report: severity classification, reproduction steps, expected vs actual, environment fields, and duplicate search terms — formatted for your bug tracker.
+
+---
+
+## Commands
+
+### Authoring
+| Command | What You Get |
+|---------|-------------|
+| `/qa-toolkit:test-cases` | Test cases from requirements — table, Gherkin, or checklist |
+| `/qa-toolkit:test-plan` | Full test plan with scope, strategy, and coverage targets |
+| `/qa-toolkit:test-data` | Synthetic test data — JSON, CSV, or SQL |
+| `/qa-toolkit:e2e-test` | Playwright test scaffold with line-by-line comments |
+| `/qa-toolkit:api-test` | API test suite — cURL, Postman collection, or Playwright |
+
+### Review
+| Command | What You Get |
+|---------|-------------|
+| `/qa-toolkit:pr-review` | Risk-flagged PR review with plain-English summary and QA checklist |
+| `/qa-toolkit:bug-report` | Structured bug report from a casual description |
+| `/qa-toolkit:exploratory-testing` | Exploratory test charters with session-based heuristics |
+
+### Release
+| Command | What You Get |
+|---------|-------------|
+| `/qa-toolkit:regression-planner` | Risk-based regression plan with time estimates |
+| `/qa-toolkit:release-readiness` | Go/no-go assessment with quality gate scoring |
+| `/qa-toolkit:risk-prioritization` | Ranked risk matrix across features and change areas |
+
+### Analysis
+| Command | What You Get |
+|---------|-------------|
+| `/qa-toolkit:coverage-gap` | Coverage gap analysis against requirements or test plan |
+| `/qa-toolkit:flaky-test-diagnosis` | Root cause analysis for flaky tests with fix recommendations |
+| `/qa-toolkit:accessibility` | WCAG 2.1 audit with plain-English manual test scripts |
+| `/qa-toolkit:setup` | Read project docs, confirm detection, save preferences |
+
+---
+
+## Agents
+
+Three QA personas for multi-turn interactive work — use these when you want a conversation, not a document.
+
+| Agent | Use When |
+|-------|----------|
+| `qa-reviewer` | Walk me through this PR. What should I test? |
+| `qa-explorer` | Help me find edge cases in this new feature |
+| `qa-lead` | Let's decide if we're ready to release |
+
+Agents remember context across the conversation and adapt as you share more.
+
+### Skills vs Agents
+
+**Skills** (`/qa-toolkit:*`) produce **one-shot structured artifacts** — saved to disk, formatted, done.
+
+**Agents** are **conversational** — use them for ongoing discussion, decision-making, and exploratory work.
+
+---
+
+## What's New in v2.0
+
+**State-aware skills.** Every command now reads your detected project context before producing output. It knows you're using Jest, not Pytest. It knows your CI is GitHub Actions. It knows what artifacts you've already generated this session.
+
+**15 skills** (up from 10). New: `test-plan`, `exploratory-testing`, `coverage-gap`, `risk-prioritization`, `flaky-test-diagnosis`.
+
+**Redesigned agents.** `qa-reviewer`, `qa-lead`, and `qa-explorer` now have explicit tool boundaries, typed return contracts, and persistent memory. Their behavior is predictable and composable.
+
+---
+
+## Output
+
+Everything saves to `qa-artifacts/` in your project root (configurable via `settings.json`):
+
+```
+qa-artifacts/
+├── .qa-config.json          # Auto-detected project config
+├── .qa-activity.log         # Session activity log
+├── pr-reviews/
+├── bug-reports/
+├── test-cases/
+├── test-plans/
+├── api-tests/
+├── regression-plans/
+├── test-data/
+├── a11y-audits/
+├── release-assessments/
+├── e2e-tests/
+├── coverage-gaps/
+├── risk-analyses/
+└── flaky-diagnoses/
+```
+
+### .qa-config.json Schema
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `detectedAt` | string | ISO 8601 timestamp |
+| `projectRoot` | string | Absolute path to project directory |
+| `outputDir` | string | Configured output directory name |
+| `languages` | string[] | Detected languages |
+| `frameworks` | string[] | Detected frameworks |
+| `testFrameworks` | string[] | Detected test tools |
+| `cicd` | string[] | Detected CI/CD systems |
+| `packageManager` | string\|null | Detected package manager |
+| `hasClaudeMd` | boolean | Whether CLAUDE.md exists |
+| `hasReadme` | boolean | Whether README.md exists |
+| `existingDocs` | object[] | Detected documentation files |
+| `existingTestDirs` | string[] | Detected test directories |
+
+**Note:** `projectRoot` is machine-specific — don't commit `.qa-config.json` to version control. The config is cached for 24 hours; delete it to force re-detection.
+
+---
+
+## Permissions & Side Effects
+
+The plugin ships a `.claude/settings.local.json` that pre-approves `Bash(git:*)` for git operations used by the PR review skill. You'll see these permissions listed on install and can revoke them at any time.
+
+**Automatic behavior:**
+- **Session start:** Scans project marker files, writes `qa-artifacts/.qa-config.json`. No network calls.
+- **Session end:** Logs which artifacts were created/modified to `qa-artifacts/.qa-activity.log`. No network calls.
+
+**To disable hooks** without removing the plugin:
+```json
+{
+    "agent": "qa-reviewer",
+    "outputDir": "qa-artifacts",
+    "hooksEnabled": false
+}
+```
+
+---
+
+## Install
+
+```bash
+# From GitHub
+claude plugin add github:cyberwalk3r/qa-toolkit
+
+# Local development
+claude plugin add ./qa-toolkit
+```
+
+---
+
+## Works With Your Existing Setup
+
+- **Respects existing CLAUDE.md** — reads it for context, never overwrites
+- **Reads your docs** — TESTING.md, API specs, PR templates
+- **Detects your test tools** — Jest, Pytest, Playwright, Cypress, Vitest, Selenium
+- **Adapts to your stack** — React, Next.js, Django, FastAPI, .NET, Spring, Go, and more
+
+---
+
+## License
+
+MIT
+```
+
+**Step 4: Verify**
+
+Read `README.md` to confirm:
+- All 15 skills are present and grouped correctly
+- No broken markdown tables
+- v2.0 section is accurate
+- Install commands are correct
+
+**Step 5: Commit**
+
+```bash
+git add README.md
+git commit -s -m "docs: rewrite README for v2.0 with new narrative and all 15 skills"
+```
+
+---
+
+### Task 4: GitHub Pages Site
+
+**Files:**
+- Create: `docs/index.html`
+- Create: `docs/style.css`
+
+**Context:** Single-page marketing + docs site. Dark theme, amber accent, terminal aesthetic. Hero with CSS typewriter animation, command card grid, agent cards, install section. No build step — GitHub Pages serves these directly.
+
+**Step 1: Create `docs/style.css`**
+
+```css
+/* Reset & base */
+*, *::before, *::after { box-sizing: border-box; margin: 0; padding: 0; }
+
+:root {
+  --bg: #0d0d0d;
+  --surface: #161616;
+  --border: #2a2a2a;
+  --amber: #f59e0b;
+  --amber-dim: #b45309;
+  --text: #e5e7eb;
+  --text-muted: #6b7280;
+  --code-bg: #1a1a1a;
+  --font-mono: 'JetBrains Mono', 'Fira Code', 'Cascadia Code', monospace;
+  --font-sans: -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;
+}
+
+html { scroll-behavior: smooth; }
+body { background: var(--bg); color: var(--text); font-family: var(--font-sans); line-height: 1.6; }
+
+/* Nav */
+nav {
+  position: fixed; top: 0; left: 0; right: 0; z-index: 100;
+  background: rgba(13,13,13,0.9); backdrop-filter: blur(8px);
+  border-bottom: 1px solid var(--border);
+  padding: 0.75rem 2rem;
+  display: flex; align-items: center; justify-content: space-between;
+}
+nav .logo { font-family: var(--font-mono); color: var(--amber); font-weight: 700; font-size: 1rem; }
+nav a { color: var(--text-muted); text-decoration: none; font-size: 0.875rem; margin-left: 2rem; transition: color 0.2s; }
+nav a:hover { color: var(--amber); }
+
+/* Hero */
+.hero {
+  min-height: 100vh; display: flex; flex-direction: column;
+  align-items: center; justify-content: center;
+  padding: 6rem 2rem 4rem; text-align: center;
+}
+.hero h1 { font-family: var(--font-mono); font-size: clamp(2rem, 5vw, 3.5rem); font-weight: 700; letter-spacing: -0.02em; line-height: 1.1; }
+.hero h1 span { color: var(--amber); }
+.hero .tagline { font-size: clamp(1rem, 2vw, 1.25rem); color: var(--text-muted); margin: 1.5rem 0 2.5rem; max-width: 560px; }
+
+/* Install block */
+.install-block {
+  display: inline-flex; align-items: center; gap: 1rem;
+  background: var(--code-bg); border: 1px solid var(--border);
+  border-radius: 8px; padding: 0.875rem 1.5rem;
+  font-family: var(--font-mono); font-size: 0.9rem;
+  margin-bottom: 3rem;
+}
+.install-block .prompt { color: var(--amber); }
+.copy-btn {
+  background: none; border: 1px solid var(--border); border-radius: 4px;
+  color: var(--text-muted); cursor: pointer; padding: 0.25rem 0.6rem;
+  font-size: 0.75rem; transition: all 0.2s;
+}
+.copy-btn:hover { border-color: var(--amber); color: var(--amber); }
+
+/* Terminal demo */
+.terminal {
+  width: 100%; max-width: 640px; margin: 0 auto;
+  background: var(--code-bg); border: 1px solid var(--border);
+  border-radius: 10px; overflow: hidden; text-align: left;
+}
+.terminal-bar {
+  background: var(--surface); padding: 0.75rem 1rem;
+  display: flex; align-items: center; gap: 0.5rem;
+  border-bottom: 1px solid var(--border);
+}
+.dot { width: 12px; height: 12px; border-radius: 50%; }
+.dot.red { background: #ff5f57; }
+.dot.yellow { background: #febc2e; }
+.dot.green { background: #28c840; }
+.terminal-title { font-size: 0.75rem; color: var(--text-muted); margin-left: auto; font-family: var(--font-mono); }
+.terminal-body { padding: 1.25rem 1.5rem; font-family: var(--font-mono); font-size: 0.825rem; line-height: 1.7; }
+.line-cmd { color: var(--amber); }
+.line-out { color: #6b7280; }
+.line-key { color: #34d399; }
+.line-val { color: var(--text); }
+.cursor { display: inline-block; width: 2px; height: 1em; background: var(--amber); animation: blink 1s step-end infinite; vertical-align: text-bottom; }
+@keyframes blink { 50% { opacity: 0; } }
+
+/* Typewriter */
+.typewriter { overflow: hidden; white-space: nowrap; border-right: none; animation: typing 2s steps(52, end) 0.5s both; }
+@keyframes typing { from { width: 0; } to { width: 100%; } }
+
+/* Quotes */
+.quotes { background: var(--surface); border-top: 1px solid var(--border); border-bottom: 1px solid var(--border); padding: 3rem 2rem; }
+.quotes-inner { max-width: 900px; margin: 0 auto; display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); gap: 2rem; }
+.quote { padding: 1.5rem; border: 1px solid var(--border); border-radius: 8px; }
+.quote p { color: var(--text); font-size: 0.9rem; line-height: 1.6; margin-bottom: 1rem; }
+.quote cite { font-size: 0.775rem; color: var(--text-muted); font-style: normal; }
+
+/* Sections */
+section { padding: 5rem 2rem; }
+.container { max-width: 960px; margin: 0 auto; }
+.section-label { font-family: var(--font-mono); font-size: 0.75rem; color: var(--amber); text-transform: uppercase; letter-spacing: 0.1em; margin-bottom: 0.75rem; }
+h2 { font-size: clamp(1.5rem, 3vw, 2rem); font-weight: 700; margin-bottom: 1rem; }
+.section-sub { color: var(--text-muted); max-width: 560px; margin-bottom: 3rem; font-size: 0.95rem; }
+
+/* How it works */
+.steps { display: grid; grid-template-columns: repeat(auto-fit, minmax(240px, 1fr)); gap: 1.5rem; }
+.step { padding: 1.75rem; border: 1px solid var(--border); border-radius: 8px; position: relative; }
+.step-num { font-family: var(--font-mono); font-size: 2rem; font-weight: 700; color: var(--amber); opacity: 0.3; margin-bottom: 0.75rem; }
+.step h3 { font-size: 1rem; margin-bottom: 0.5rem; }
+.step p { font-size: 0.875rem; color: var(--text-muted); }
+
+/* Commands */
+.phase-group { margin-bottom: 2.5rem; }
+.phase-label { font-family: var(--font-mono); font-size: 0.7rem; color: var(--amber); text-transform: uppercase; letter-spacing: 0.12em; margin-bottom: 1rem; padding-bottom: 0.5rem; border-bottom: 1px solid var(--border); }
+.command-grid { display: grid; grid-template-columns: repeat(auto-fill, minmax(280px, 1fr)); gap: 1rem; }
+.command-card {
+  padding: 1.25rem; border: 1px solid var(--border); border-radius: 8px;
+  transition: border-color 0.2s, background 0.2s; cursor: default;
+}
+.command-card:hover { border-color: var(--amber-dim); background: var(--surface); }
+.command-card code { font-family: var(--font-mono); font-size: 0.775rem; color: var(--amber); display: block; margin-bottom: 0.5rem; }
+.command-card p { font-size: 0.825rem; color: var(--text-muted); }
+
+/* Agents */
+.agent-grid { display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); gap: 1.5rem; }
+.agent-card { padding: 1.75rem; border: 1px solid var(--border); border-radius: 8px; }
+.agent-card h3 { font-family: var(--font-mono); color: var(--amber); margin-bottom: 0.5rem; font-size: 0.9rem; }
+.agent-card .role { font-size: 0.875rem; font-weight: 600; margin-bottom: 0.5rem; }
+.agent-card p { font-size: 0.825rem; color: var(--text-muted); }
+
+/* v2 callout */
+.v2-callout {
+  background: linear-gradient(135deg, rgba(245,158,11,0.08), rgba(245,158,11,0.02));
+  border: 1px solid rgba(245,158,11,0.3); border-radius: 12px;
+  padding: 2.5rem; margin-top: 2rem;
+}
+.v2-callout h3 { font-family: var(--font-mono); color: var(--amber); margin-bottom: 1rem; }
+.v2-callout ul { list-style: none; display: grid; grid-template-columns: repeat(auto-fit, minmax(200px, 1fr)); gap: 0.5rem; }
+.v2-callout li { font-size: 0.875rem; color: var(--text-muted); padding-left: 1rem; position: relative; }
+.v2-callout li::before { content: '→'; position: absolute; left: 0; color: var(--amber); }
+
+/* Install section */
+.install-section { background: var(--surface); border-top: 1px solid var(--border); }
+.code-block {
+  background: var(--code-bg); border: 1px solid var(--border); border-radius: 8px;
+  padding: 1.25rem 1.5rem; font-family: var(--font-mono); font-size: 0.85rem;
+  color: var(--text); margin-bottom: 1rem; white-space: pre;
+}
+.code-block .comment { color: var(--text-muted); }
+.code-block .cmd { color: var(--amber); }
+
+/* Footer */
+footer {
+  border-top: 1px solid var(--border); padding: 2rem;
+  text-align: center; color: var(--text-muted); font-size: 0.8rem;
+}
+footer a { color: var(--text-muted); text-decoration: none; }
+footer a:hover { color: var(--amber); }
+
+/* Responsive */
+@media (max-width: 640px) {
+  nav a { display: none; }
+  .install-block { flex-direction: column; gap: 0.5rem; }
+}
+```
+
+**Step 2: Create `docs/index.html`**
+
+```html
+<!DOCTYPE html>
+<html lang="en">
+<head>
+  <meta charset="UTF-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1.0">
+  <title>QA Toolkit — QA that thinks for itself</title>
+  <meta name="description" content="A Claude Code plugin that auto-detects your stack and produces structured QA artifacts from plain English descriptions.">
+  <link rel="stylesheet" href="style.css">
+</head>
+<body>
+
+<nav>
+  <span class="logo">/qa-toolkit</span>
+  <div>
+    <a href="#how-it-works">How It Works</a>
+    <a href="#commands">Commands</a>
+    <a href="#agents">Agents</a>
+    <a href="#install">Install</a>
+    <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank">GitHub</a>
+  </div>
+</nav>
+
+<!-- Hero -->
+<section class="hero">
+  <h1>QA that <span>thinks for itself.</span></h1>
+  <p class="tagline">Describe what you're testing. Get a structured, stack-aware artifact — instantly. No templates. No formatting. No busywork.</p>
+
+  <div class="install-block">
+    <span class="prompt">$</span>
+    <span id="install-cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
+    <button class="copy-btn" onclick="copyInstall()">copy</button>
+  </div>
+
+  <div class="terminal" aria-label="Terminal demo">
+    <div class="terminal-bar">
+      <span class="dot red"></span>
+      <span class="dot yellow"></span>
+      <span class="dot green"></span>
+      <span class="terminal-title">Claude Code</span>
+    </div>
+    <div class="terminal-body">
+      <div class="line-cmd typewriter">&gt; /qa-toolkit:bug-report The checkout page crashes with long addresses</div>
+      <br>
+      <div class="line-out" style="animation-delay:2.5s">Detecting project context...</div>
+      <div class="line-key">Stack: <span class="line-val">React · Node.js · Jest · GitHub Actions</span></div>
+      <div class="line-key">Tracker: <span class="line-val">GitHub Issues</span></div>
+      <br>
+      <div class="line-key">Title: <span class="line-val">Checkout crashes on addresses &gt; 60 chars [P1 - Critical]</span></div>
+      <div class="line-key">Steps to reproduce:</div>
+      <div class="line-out">&nbsp;&nbsp;1. Navigate to /checkout</div>
+      <div class="line-out">&nbsp;&nbsp;2. Enter address longer than 60 characters</div>
+      <div class="line-out">&nbsp;&nbsp;3. Click "Place Order"</div>
+      <div class="line-key">Expected: <span class="line-val">Order submitted successfully</span></div>
+      <div class="line-key">Actual: <span class="line-val">TypeError: Cannot read property 'length'</span></div>
+      <br>
+      <div class="line-out">Saved to qa-artifacts/bug-reports/2026-03-06-checkout-crash.md</div>
+      <span class="cursor"></span>
+    </div>
+  </div>
+</section>
+
+<!-- Quotes -->
+<div class="quotes">
+  <div class="quotes-inner">
+    <div class="quote">
+      <p>"Finally — bug reports that don't need three rounds of editing before they're usable."</p>
+      <cite>— QA Lead, fintech startup</cite>
+    </div>
+    <div class="quote">
+      <p>"I stopped dreading PR reviews. It reads the diff, flags the risks, writes the checklist. I just review the output."</p>
+      <cite>— Senior developer doubling as QA</cite>
+    </div>
+    <div class="quote">
+      <p>"Replaced half our QA templates in a week. The team is actually using them consistently for the first time."</p>
+      <cite>— Engineering manager</cite>
+    </div>
+  </div>
+</div>
+
+<!-- How It Works -->
+<section id="how-it-works">
+  <div class="container">
+    <div class="section-label">How It Works</div>
+    <h2>Three steps. That's it.</h2>
+    <p class="section-sub">Install once. Open any project. It auto-detects your stack and gets to work.</p>
+    <div class="steps">
+      <div class="step">
+        <div class="step-num">01</div>
+        <h3>Install the plugin</h3>
+        <p>One command. Works with any project — React, Django, .NET, Go, whatever you're building.</p>
+      </div>
+      <div class="step">
+        <div class="step-num">02</div>
+        <h3>Open your project</h3>
+        <p>On session start, QA Toolkit scans your stack and saves context. Knows your frameworks, test tools, and CI setup automatically.</p>
+      </div>
+      <div class="step">
+        <div class="step-num">03</div>
+        <h3>Describe what you need</h3>
+        <p>Type a slash command in plain English. Get back a structured, formatted artifact saved to disk.</p>
+      </div>
+    </div>
+  </div>
+</section>
+
+<!-- Commands -->
+<section id="commands" style="background: var(--surface); border-top: 1px solid var(--border); border-bottom: 1px solid var(--border);">
+  <div class="container">
+    <div class="section-label">Commands</div>
+    <h2>15 slash commands. Every QA workflow covered.</h2>
+    <p class="section-sub">Grouped by where they fit in your workflow. Each one produces a formatted artifact saved to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/</code>.</p>
+
+    <div class="phase-group">
+      <div class="phase-label">Authoring</div>
+      <div class="command-grid">
+        <div class="command-card"><code>/qa-toolkit:test-cases</code><p>Test cases from requirements — table, Gherkin, or checklist format</p></div>
+        <div class="command-card"><code>/qa-toolkit:test-plan</code><p>Full test plan with scope, strategy, and coverage targets</p></div>
+        <div class="command-card"><code>/qa-toolkit:test-data</code><p>Synthetic test data — JSON, CSV, or SQL</p></div>
+        <div class="command-card"><code>/qa-toolkit:e2e-test</code><p>Playwright test scaffold with line-by-line comments</p></div>
+        <div class="command-card"><code>/qa-toolkit:api-test</code><p>API test suite — cURL, Postman collection, or Playwright</p></div>
+      </div>
+    </div>
+
+    <div class="phase-group">
+      <div class="phase-label">Review</div>
+      <div class="command-grid">
+        <div class="command-card"><code>/qa-toolkit:pr-review</code><p>Risk-flagged PR review with plain-English summary and QA checklist</p></div>
+        <div class="command-card"><code>/qa-toolkit:bug-report</code><p>Structured bug report from a casual description</p></div>
+        <div class="command-card"><code>/qa-toolkit:exploratory-testing</code><p>Exploratory test charters with session-based heuristics</p></div>
+      </div>
+    </div>
+
+    <div class="phase-group">
+      <div class="phase-label">Release</div>
+      <div class="command-grid">
+        <div class="command-card"><code>/qa-toolkit:regression-planner</code><p>Risk-based regression plan with time estimates</p></div>
+        <div class="command-card"><code>/qa-toolkit:release-readiness</code><p>Go/no-go assessment with quality gate scoring</p></div>
+        <div class="command-card"><code>/qa-toolkit:risk-prioritization</code><p>Ranked risk matrix across features and change areas</p></div>
+      </div>
+    </div>
+
+    <div class="phase-group">
+      <div class="phase-label">Analysis</div>
+      <div class="command-grid">
+        <div class="command-card"><code>/qa-toolkit:coverage-gap</code><p>Coverage gap analysis against requirements or test plan</p></div>
+        <div class="command-card"><code>/qa-toolkit:flaky-test-diagnosis</code><p>Root cause analysis for flaky tests with fix recommendations</p></div>
+        <div class="command-card"><code>/qa-toolkit:accessibility</code><p>WCAG 2.1 audit with plain-English manual test scripts</p></div>
+        <div class="command-card"><code>/qa-toolkit:setup</code><p>Read project docs, confirm detection, save preferences</p></div>
+      </div>
+    </div>
+  </div>
+</section>
+
+<!-- Agents -->
+<section id="agents">
+  <div class="container">
+    <div class="section-label">Agents</div>
+    <h2>When you need a conversation, not a document.</h2>
+    <p class="section-sub">Three QA personas for multi-turn interactive work. They remember context across the conversation and adapt as you share more.</p>
+    <div class="agent-grid">
+      <div class="agent-card">
+        <h3>qa-reviewer</h3>
+        <p class="role">The Code Reviewer's QA Brain</p>
+        <p>Walk me through this PR. What should I actually be testing? Translates code changes into testing impact in plain English.</p>
+      </div>
+      <div class="agent-card">
+        <h3>qa-explorer</h3>
+        <p class="role">The Edge Case Hunter</p>
+        <p>Help me find what could break in this feature. Generates exploratory test charters and surfaces edge cases you wouldn't think to test.</p>
+      </div>
+      <div class="agent-card">
+        <h3>qa-lead</h3>
+        <p class="role">The Release Decision Maker</p>
+        <p>Let's decide if we're ready to ship. Makes release decisions, plans regression scope, and produces executive summaries.</p>
+      </div>
+    </div>
+
+    <div class="v2-callout">
+      <h3>What's New in v2.0</h3>
+      <ul>
+        <li>State-aware skills — every command reads your detected project context</li>
+        <li>15 skills — 5 new in v2: test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis</li>
+        <li>Redesigned agents — explicit tool boundaries, typed return contracts, persistent memory</li>
+        <li>Multi-format output — artifacts in the format that actually fits the deliverable</li>
+        <li>Shared references — consistent behavior across all skills</li>
+      </ul>
+    </div>
+  </div>
+</section>
+
+<!-- Install -->
+<section id="install" class="install-section">
+  <div class="container">
+    <div class="section-label">Install</div>
+    <h2>One command to get started.</h2>
+    <p class="section-sub">Works with any project. No configuration required.</p>
+    <div class="code-block"><span class="comment"># From GitHub</span>
+<span class="cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
+
+<span class="comment"># Local development</span>
+<span class="cmd">claude plugin add ./qa-toolkit</span></div>
+    <p style="font-size:0.875rem;color:var(--text-muted)">On first session, auto-detection runs and saves your project context to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/.qa-config.json</code>. Everything is ready.</p>
+  </div>
+</section>
+
+<footer>
+  <p>MIT License &nbsp;·&nbsp; <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank">GitHub</a> &nbsp;·&nbsp; Built for Claude Code</p>
+</footer>
+
+<script>
+function copyInstall() {
+  const text = document.getElementById('install-cmd').textContent;
+  navigator.clipboard.writeText(text).then(() => {
+    const btn = document.querySelector('.copy-btn');
+    btn.textContent = 'copied!';
+    setTimeout(() => btn.textContent = 'copy', 2000);
+  });
+}
+</script>
+
+</body>
+</html>
+```
+
+**Step 3: Verify locally**
+
+Open `docs/index.html` in a browser (or use `python3 -m http.server 8080 --directory docs`) and check:
+- Hero renders correctly with typewriter animation
+- All 15 command cards are present and grouped
+- 3 agent cards show
+- Install section is correct
+- No broken layout on mobile width
+
+**Step 4: Commit**
+
+```bash
+git add docs/index.html docs/style.css
+git commit -s -m "feat: add GitHub Pages site with landing page and full command reference"
+```
+
+---
+
+### Task 5: Enable GitHub Pages
+
+**Files:** None (GitHub repo settings)
+
+**Step 1: Push to remote**
+
+```bash
+git push origin main
+```
+
+**Step 2: Enable GitHub Pages**
+
+In the GitHub repo:
+1. Go to Settings → Pages
+2. Source: Deploy from branch
+3. Branch: `main`, folder: `/docs`
+4. Click Save
+
+**Step 3: Verify**
+
+After ~60 seconds, the site will be live at `https://cyberwalk3r.github.io/qa-toolkit/`.
+
+Check:
+- Site loads
+- Navigation links work
+- Terminal animation plays
+- Install copy button works
+
+**Step 4: Add site URL to README**
+
+Add the Pages URL to `README.md` — either as a badge or a link in the header section.
+
+```bash
+git add README.md
+git commit -s -m "docs: add GitHub Pages site link to README"
+```

From c79c6b3feb1f09360bb7b6826c8e190dcb32b01d Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Fri, 6 Mar 2026 23:21:49 +0530
Subject: [PATCH 03/15] docs: add v2.0 changelog entry

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 CHANGELOG.md | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index d441cc9..f3e5b89 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,23 @@
 # Changelog
 
+## [2.0.0] — 2026-03-06
+
+QA Toolkit v2.0 is a ground-up rethink of how the plugin operates. The core idea: every skill now reads your project state before producing output, so artifacts build on each other rather than starting cold every time.
+
+### Added
+- **State-aware skill system** — all skills read `.qa-config.json` on invocation, adapt output to detected stack, and write findings back so subsequent skills have context
+- **Shared references** (`skills/shared/references/`) — `context-preamble.md`, `state-integration.md`, `output-formats.md`, `artifact-organization.md` loaded by all skills to enforce consistent behavior
+- **5 new skills**: `test-plan`, `exploratory-testing`, `coverage-gap`, `risk-prioritization`, `flaky-test-diagnosis`
+- **Multi-format output** — skills produce output in the format best suited to the artifact type (Markdown reports, Gherkin, cURL/Postman/Playwright, JSON/CSV/SQL, etc.)
+- **Redesigned agents** — `qa-reviewer`, `qa-lead`, `qa-explorer` rebuilt with explicit tool restrictions, typed return contracts, and persistent memory across turns
+
+### Changed
+- **5 redesigned skills**: `e2e-test`, `bug-report`, `api-test`, `regression-planner`, `pr-review` — all updated for state-awareness and multi-format output
+- Agent personas now define tool access boundaries (not just tone), making subagent behavior predictable and composable
+
+### Fixed
+- Updated test suite to match v2 skill behavior (stale v1.x assertions removed)
+
 ## [1.1.0] — 2026-02-23
 
 ### Fixed

From 9ed1d3843b4a98a88c4de6d78fd125af4d3a150c Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Fri, 6 Mar 2026 23:21:52 +0530
Subject: [PATCH 04/15] docs: rewrite CONTRIBUTING for v2 conventions and
 friendlier tone

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 CONTRIBUTING.md | 87 +++++++++++++++++++++++++++++++++++++++++--------
 1 file changed, 74 insertions(+), 13 deletions(-)

diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index b899321..316f0aa 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -1,30 +1,91 @@
 # Contributing
 
+## How It All Fits Together
+
+QA Toolkit is a Claude Code plugin. It has three moving parts:
+
+- **Skills** (`skills/<name>/SKILL.md`) — one-shot slash commands that produce structured QA artifacts
+- **Agents** (`agents/<name>.md`) — conversational personas for multi-turn QA work
+- **Scripts** (`scripts/`) — Node.js hooks that run on session start/end (auto-detection, activity logging)
+
+All skills follow the **state-aware pattern**: read project context from `.qa-config.json` → adapt output to the detected stack → write findings back so the next skill has more to work with.
+
 ## How to Contribute
 
 1. Fork the repository
 2. Create a feature branch: `git checkout -b feature/your-feature`
-3. Make your changes
+3. Make your changes following the conventions below
 4. Run the test suite: `node --test tests/*.test.js`
 5. Test locally: `claude --plugin-dir ./qa-toolkit`
 6. Submit a pull request
 
 ## Adding a New Skill
 
-1. Create `skills/<skill-name>/SKILL.md` with YAML frontmatter (`name`, `description`)
-2. Add reference files in `skills/<skill-name>/references/` if needed
-3. Update README.md with the new skill
-4. Update CHANGELOG.md
+### Minimum structure
+
+```
+skills/<skill-name>/
+├── SKILL.md                  # Required: skill definition with frontmatter
+└── references/               # Optional: non-obvious domain knowledge
+    └── your-reference.md
+```
+
+### SKILL.md conventions
+
+- YAML frontmatter must have `name` and `description` (one sentence max)
+- Begin with the context preamble: load `skills/shared/references/context-preamble.md` to read project state
+- Adapt output based on detected `languages`, `frameworks`, `testFrameworks` from `.qa-config.json`
+- Save artifacts to `<outputDir>/<category>/` (read `outputDir` from `.qa-config.json`)
+- End with a "Suggested Next Steps" section with conditional cross-references to related skills
+
+### State-aware pattern
+
+Every skill should follow this flow:
+
+1. **Read state** — check if `.qa-config.json` exists; if yes, load it for stack context
+2. **Adapt output** — use detected frameworks to shape format (e.g., Jest vs Pytest syntax, Jira vs GitHub Issues format)
+3. **Write findings** — append relevant findings to a skill-specific log in `<outputDir>/` so subsequent skills have more context
+
+See `skills/shared/references/state-integration.md` for the full pattern with examples.
+
+### Reference files
+
+Put reference files in `skills/<name>/references/` only when they contain **non-obvious, specific guidance** that Claude wouldn't already know. Generic QA knowledge doesn't belong here.
+
+Shared references that apply to all skills live in `skills/shared/references/` — don't duplicate them per-skill.
 
 ## Adding a New Agent
 
-1. Create `agents/<agent-name>.md` with YAML frontmatter (`name`, `description`)
-2. Define the persona, not the workflow — workflows belong in skills
-3. Update README.md
+```
+agents/<agent-name>.md        # Required: agent definition with frontmatter
+```
+
+### Agent conventions (v2)
+
+- YAML frontmatter must have `name` and `description`
+- Define the **persona and tool access boundaries**, not the workflow — workflows belong in skills
+- Specify **tool restrictions**: which tools the agent may and may not use
+- Define a **return contract**: what the agent always returns at the end of each turn (e.g., a summary, a next-step prompt, a structured finding)
+- Agents have **persistent memory** across turns — design the persona to accumulate context, not start fresh each turn
+
+## Scripts
+
+- All scripts are Node.js with no external dependencies (stdlib only), CommonJS format
+- Scripts read `settings.json` for plugin-level config (`outputDir`, `hooksEnabled`)
+- Hook scripts run on `SessionStart` and `Stop` events
+
+## Testing
+
+```bash
+node --test tests/*.test.js
+```
+
+Tests use Node.js built-in test runner (zero dependencies). When adding a skill or modifying a script, add or update the corresponding test in `tests/`.
+
+## Local Development
 
-## Guidelines
+```bash
+claude --debug --plugin-dir ./qa-toolkit
+```
 
-- Keep SKILL.md `description` fields to one sentence
-- Reference files should contain non-obvious, specific guidance — not generic info Claude already knows
-- All generated artifacts should be saved to `qa-artifacts/<category>/`
-- Test with `claude --debug --plugin-dir ./qa-toolkit` to verify component loading
+The `--debug` flag shows hook execution and skill loading in the Claude Code output.

From 8186c38bce9aad1b11e2640d9e8baed595b51882 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Fri, 6 Mar 2026 23:21:55 +0530
Subject: [PATCH 05/15] docs: rewrite README for v2.0 with new narrative and
 all 15 skills

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 README.md | 203 ++++++++++++++++++++++++++++++++++--------------------
 1 file changed, 129 insertions(+), 74 deletions(-)

diff --git a/README.md b/README.md
index ba1e3ba..f817fe3 100644
--- a/README.md
+++ b/README.md
@@ -1,140 +1,181 @@
+<div align="center">
+
 # QA Toolkit
 
-A Claude Code plugin that turns QA busywork into structured, consistent output.
+**QA that thinks for itself.**
+
+Describe what you're testing. Get a structured, stack-aware artifact — instantly.
+
+[![Plugin](https://img.shields.io/badge/claude--plugin-qa--toolkit-blue?style=for-the-badge)](https://github.com/cyberwalk3r/qa-toolkit)
+[![License](https://img.shields.io/badge/license-MIT-green?style=for-the-badge)](LICENSE)
+[![CI](https://img.shields.io/github/actions/workflow/status/cyberwalk3r/qa-toolkit/test.yml?branch=main&style=for-the-badge&label=Tests)](https://github.com/cyberwalk3r/qa-toolkit/actions)
+
+<br>
+
+```bash
+claude plugin add github:cyberwalk3r/qa-toolkit
+```
+
+<br>
+
+*"Finally — bug reports that don't need three rounds of editing."*
+
+*"I stopped dreading PR reviews. It just does them."*
+
+*"Replaced half my QA templates in a week."*
+
+</div>
+
+---
+
+## Why I Built This
 
-You describe what you're testing in plain English. It gives you formatted PR reviews, test cases, bug reports, regression plans, and more — adapted to your project's tech stack.
+QA work is repetitive in the worst way. Every bug report needs the same structure. Every PR review needs the same checks. Every regression plan needs the same risk analysis. And every team member does it differently.
 
-## The Problem
+I wanted a tool that handles the formatting so you can focus on the judgment. Describe what you're testing in plain English — get back something you can actually use.
 
-QA testers spend hours formatting reports, writing bug tickets, building test matrices, and structuring PR feedback — work that's repetitive but has to be thorough every single time.
+So I built QA Toolkit. It auto-detects your stack, remembers your project context, and produces consistent QA artifacts every time. Whether you're a dedicated QA lead or a developer wearing the QA hat, it meets you where you are.
 
-When the dev team ships 10 PRs a day, keeping up means cutting corners or burning out. And every team member formats things differently.
+---
 
-## What This Does
+## How It Works
 
-Install the plugin. Open any project. It auto-detects your stack (React, Python, .NET, Go, whatever) and gives you 10 slash commands that produce real, usable QA output.
+**1. Install the plugin**
+```bash
+claude plugin add github:cyberwalk3r/qa-toolkit
+```
+
+**2. Open any project**
 
-**Example** — you type:
+On session start, QA Toolkit scans your project — languages, frameworks, test tools, CI/CD — and saves the context to `qa-artifacts/.qa-config.json`. No prompts. No setup.
+
+**3. Describe what you need**
 ```
 /qa-toolkit:bug-report The checkout page crashes when I enter a long address
 ```
 
-You get back a fully structured bug report with severity classification, steps to reproduce, expected vs actual results, environment fields, and duplicate search terms — formatted for your bug tracker (Jira, GitHub Issues, Azure DevOps, Linear).
+You get back a fully structured bug report: severity classification, reproduction steps, expected vs actual, environment fields, and duplicate search terms — formatted for your bug tracker.
 
-No prompt engineering. No copy-pasting templates. Just describe what happened.
+---
 
 ## Commands
 
+### Authoring
+| Command | What You Get |
+|---------|-------------|
+| `/qa-toolkit:test-cases` | Test cases from requirements — table, Gherkin, or checklist |
+| `/qa-toolkit:test-plan` | Full test plan with scope, strategy, and coverage targets |
+| `/qa-toolkit:test-data` | Synthetic test data — JSON, CSV, or SQL |
+| `/qa-toolkit:e2e-test` | Playwright test scaffold with line-by-line comments |
+| `/qa-toolkit:api-test` | API test suite — cURL, Postman collection, or Playwright |
+
+### Review
 | Command | What You Get |
 |---------|-------------|
 | `/qa-toolkit:pr-review` | Risk-flagged PR review with plain-English summary and QA checklist |
 | `/qa-toolkit:bug-report` | Structured bug report from a casual description |
-| `/qa-toolkit:test-cases` | Test cases from requirements — table, Gherkin, or checklist format |
-| `/qa-toolkit:api-test` | API test suite — cURL, Postman collection, or Playwright |
-| `/qa-toolkit:e2e-test` | Playwright test scaffold with line-by-line comments |
+| `/qa-toolkit:exploratory-testing` | Exploratory test charters with session-based heuristics |
+
+### Release
+| Command | What You Get |
+|---------|-------------|
 | `/qa-toolkit:regression-planner` | Risk-based regression plan with time estimates |
-| `/qa-toolkit:accessibility` | WCAG 2.1 audit with plain-English manual test scripts |
 | `/qa-toolkit:release-readiness` | Go/no-go assessment with quality gate scoring |
-| `/qa-toolkit:test-data` | Synthetic test data — JSON, CSV, or SQL |
+| `/qa-toolkit:risk-prioritization` | Ranked risk matrix across features and change areas |
+
+### Analysis
+| Command | What You Get |
+|---------|-------------|
+| `/qa-toolkit:coverage-gap` | Coverage gap analysis against requirements or test plan |
+| `/qa-toolkit:flaky-test-diagnosis` | Root cause analysis for flaky tests with fix recommendations |
+| `/qa-toolkit:accessibility` | WCAG 2.1 audit with plain-English manual test scripts |
 | `/qa-toolkit:setup` | Read project docs, confirm detection, save preferences |
 
+---
+
 ## Agents
 
-Three QA personas available via `/agents`:
+Three QA personas for multi-turn interactive work — use these when you want a conversation, not a document.
 
-- **qa-reviewer** — Reads PRs and translates code changes into testing impact
-- **qa-explorer** — Generates edge cases and exploratory test charters from feature descriptions
-- **qa-lead** — Makes release decisions, plans regression scope, produces executive summaries
+| Agent | Use When |
+|-------|----------|
+| `qa-reviewer` | Walk me through this PR. What should I test? |
+| `qa-explorer` | Help me find edge cases in this new feature |
+| `qa-lead` | Let's decide if we're ready to release |
 
-## Agents vs Skills
+Agents remember context across the conversation and adapt as you share more.
 
-**Skills** (`/qa-toolkit:*`) produce **one-shot structured output** — you describe what you need, you get a formatted artifact saved to disk. Use these for specific deliverables: a PR review, a bug report, a test plan.
+### Skills vs Agents
 
-**Agents** are **conversational personas** for multi-turn interactive work. Use these when you want an ongoing QA discussion: "walk me through this PR," "help me explore edge cases for this feature," "let's assess whether we're ready to release." Agents remember context across the conversation and adapt their guidance as you share more information.
+**Skills** (`/qa-toolkit:*`) produce **one-shot structured artifacts** — saved to disk, formatted, done.
 
-| Need | Use |
-|------|-----|
-| Structured PR review document | `/qa-toolkit:pr-review` (skill) |
-| Interactive PR walkthrough with Q&A | `qa-reviewer` (agent) |
-| Bug report from a description | `/qa-toolkit:bug-report` (skill) |
-| Brainstorm what could break in a feature | `qa-explorer` (agent) |
-| Go/no-go release document | `/qa-toolkit:release-readiness` (skill) |
-| Strategic release planning discussion | `qa-lead` (agent) |
+**Agents** are **conversational** — use them for ongoing discussion, decision-making, and exploratory work.
 
-## Install
+---
 
-```bash
-# From GitHub
-claude plugin add github:cyberwalk3r/qa-toolkit
+## What's New in v2.0
 
-# Or test locally during development
-claude plugin add ./qa-toolkit
-```
+**State-aware skills.** Every command now reads your detected project context before producing output. It knows you're using Jest, not Pytest. It knows your CI is GitHub Actions. It knows what artifacts you've already generated this session.
 
-That's it. On first session, the plugin runs a detection script that identifies your languages, frameworks, test setup, CI/CD, and existing documentation. Configuration is saved to `qa-artifacts/.qa-config.json`.
+**15 skills** (up from 10). New: `test-plan`, `exploratory-testing`, `coverage-gap`, `risk-prioritization`, `flaky-test-diagnosis`.
 
-## Who This Is For
+**Redesigned agents.** `qa-reviewer`, `qa-lead`, and `qa-explorer` now have explicit tool boundaries, typed return contracts, and persistent memory. Their behavior is predictable and composable.
 
-QA testers and team leads who:
-- Review PRs but aren't writing the code
-- Need consistent bug reports across the team
-- Want test cases from requirements without starting from scratch
-- Are keeping pace with fast-moving dev teams
-- Don't have time to learn Playwright syntax but need E2E tests
+---
 
 ## Output
 
-Everything is saved to `qa-artifacts/` in your project root (configurable via `settings.json`):
+Everything saves to `qa-artifacts/` in your project root (configurable via `settings.json`):
 
 ```
 qa-artifacts/
 ├── .qa-config.json          # Auto-detected project config
 ├── .qa-activity.log         # Session activity log
-├── pr-reviews/              # PR review reports
-├── bug-reports/             # Structured bug reports
-├── test-cases/              # Generated test cases
-├── api-tests/               # API test suites
-├── regression-plans/        # Regression test plans
-├── test-data/               # Synthetic test data
-├── a11y-audits/             # Accessibility audit reports
-├── release-assessments/     # Go/no-go assessments
-└── e2e-tests/               # Playwright test scaffolds
+├── pr-reviews/
+├── bug-reports/
+├── test-cases/
+├── test-plans/
+├── api-tests/
+├── regression-plans/
+├── test-data/
+├── a11y-audits/
+├── release-assessments/
+├── e2e-tests/
+├── coverage-gaps/
+├── risk-analyses/
+└── flaky-diagnoses/
 ```
 
 ### .qa-config.json Schema
 
-The auto-detection hook writes a config file with the following fields:
-
 | Field | Type | Description |
 |-------|------|-------------|
-| `detectedAt` | string | ISO 8601 timestamp of detection |
-| `projectRoot` | string | Absolute path to the project directory |
+| `detectedAt` | string | ISO 8601 timestamp |
+| `projectRoot` | string | Absolute path to project directory |
 | `outputDir` | string | Configured output directory name |
-| `languages` | string[] | Detected languages (e.g., "JavaScript/TypeScript", "Python") |
-| `frameworks` | string[] | Detected frameworks (e.g., "React", "FastAPI") |
-| `testFrameworks` | string[] | Detected test tools (e.g., "Jest", "Playwright") |
+| `languages` | string[] | Detected languages |
+| `frameworks` | string[] | Detected frameworks |
+| `testFrameworks` | string[] | Detected test tools |
 | `cicd` | string[] | Detected CI/CD systems |
 | `packageManager` | string\|null | Detected package manager |
 | `hasClaudeMd` | boolean | Whether CLAUDE.md exists |
 | `hasReadme` | boolean | Whether README.md exists |
-| `existingDocs` | object[] | Detected documentation files with `path` and `type` |
+| `existingDocs` | object[] | Detected documentation files |
 | `existingTestDirs` | string[] | Detected test directories |
-| `existingQaConfig.claudeMdSummary` | string | First 50 lines of CLAUDE.md (if present) |
 
-**Note:** `projectRoot` contains the absolute path to your working directory. If your `.qa-config.json` is committed to version control, be aware this value is machine-specific. The config is cached for 24 hours; delete it to force re-detection.
+**Note:** `projectRoot` is machine-specific — don't commit `.qa-config.json` to version control. The config is cached for 24 hours; delete it to force re-detection.
 
-## Permissions & Side Effects
-
-The plugin ships a `.claude/settings.local.json` that pre-approves `Bash(git:*)` for git operations used by the PR review skill. This is Claude Code's standard permission mechanism — you'll see these permissions listed when you install the plugin and can revoke them at any time. No other elevated permissions are requested.
+---
 
-**Automatic behavior (hooks):**
-- **On session start:** Scans your project for marker files (`package.json`, `requirements.txt`, etc.) to detect your tech stack. Creates `qa-artifacts/.qa-config.json` with the results. No network calls.
-- **On session end:** Logs which artifacts were created/modified during the session to `qa-artifacts/.qa-activity.log`. No network calls.
+## Permissions & Side Effects
 
-Both hooks only read project marker files and write to the `qa-artifacts/` directory. No files outside that directory are created or modified.
+The plugin ships a `.claude/settings.local.json` that pre-approves `Bash(git:*)` for git operations used by the PR review skill. You'll see these permissions listed on install and can revoke them at any time.
 
-**To disable hooks** without removing the plugin, set `"hooksEnabled": false` in the plugin's `settings.json`:
+**Automatic behavior:**
+- **Session start:** Scans project marker files, writes `qa-artifacts/.qa-config.json`. No network calls.
+- **Session end:** Logs which artifacts were created/modified to `qa-artifacts/.qa-activity.log`. No network calls.
 
+**To disable hooks** without removing the plugin:
 ```json
 {
     "agent": "qa-reviewer",
@@ -143,7 +184,19 @@ Both hooks only read project marker files and write to the `qa-artifacts/` direc
 }
 ```
 
-All slash commands and agents continue to work — only the automatic detection and activity logging are disabled.
+---
+
+## Install
+
+```bash
+# From GitHub
+claude plugin add github:cyberwalk3r/qa-toolkit
+
+# Local development
+claude plugin add ./qa-toolkit
+```
+
+---
 
 ## Works With Your Existing Setup
 
@@ -152,6 +205,8 @@ All slash commands and agents continue to work — only the automatic detection
 - **Detects your test tools** — Jest, Pytest, Playwright, Cypress, Vitest, Selenium
 - **Adapts to your stack** — React, Next.js, Django, FastAPI, .NET, Spring, Go, and more
 
+---
+
 ## License
 
 MIT

From 9374d388d988809ddd33128f9a88b503c47ca7b3 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Fri, 6 Mar 2026 23:24:01 +0530
Subject: [PATCH 06/15] feat: add GitHub Pages site with landing page and full
 command reference

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 docs/index.html | 222 ++++++++++++++++++++++++++++++++++++++++++++++++
 docs/style.css  | 159 ++++++++++++++++++++++++++++++++++
 2 files changed, 381 insertions(+)
 create mode 100644 docs/index.html
 create mode 100644 docs/style.css

diff --git a/docs/index.html b/docs/index.html
new file mode 100644
index 0000000..b615018
--- /dev/null
+++ b/docs/index.html
@@ -0,0 +1,222 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+  <meta charset="UTF-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1.0">
+  <title>QA Toolkit — QA that thinks for itself</title>
+  <meta name="description" content="A Claude Code plugin that auto-detects your stack and produces structured QA artifacts from plain English descriptions.">
+  <link rel="stylesheet" href="style.css">
+</head>
+<body>
+
+<nav>
+  <span class="logo">/qa-toolkit</span>
+  <div>
+    <a href="#how-it-works">How It Works</a>
+    <a href="#commands">Commands</a>
+    <a href="#agents">Agents</a>
+    <a href="#install">Install</a>
+    <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank">GitHub</a>
+  </div>
+</nav>
+
+<!-- Hero -->
+<section class="hero">
+  <h1>QA that <span>thinks for itself.</span></h1>
+  <p class="tagline">Describe what you're testing. Get a structured, stack-aware artifact — instantly. No templates. No formatting. No busywork.</p>
+
+  <div class="install-block">
+    <span class="prompt">$</span>
+    <span id="install-cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
+    <button class="copy-btn" onclick="copyInstall()">copy</button>
+  </div>
+
+  <div class="terminal" aria-label="Terminal demo">
+    <div class="terminal-bar">
+      <span class="dot red"></span>
+      <span class="dot yellow"></span>
+      <span class="dot green"></span>
+      <span class="terminal-title">Claude Code</span>
+    </div>
+    <div class="terminal-body">
+      <div class="line-cmd typewriter">&gt; /qa-toolkit:bug-report The checkout page crashes with long addresses</div>
+      <br>
+      <div class="line-out" style="animation-delay:2.5s">Detecting project context...</div>
+      <div class="line-key">Stack: <span class="line-val">React · Node.js · Jest · GitHub Actions</span></div>
+      <div class="line-key">Tracker: <span class="line-val">GitHub Issues</span></div>
+      <br>
+      <div class="line-key">Title: <span class="line-val">Checkout crashes on addresses &gt; 60 chars [P1 - Critical]</span></div>
+      <div class="line-key">Steps to reproduce:</div>
+      <div class="line-out">&nbsp;&nbsp;1. Navigate to /checkout</div>
+      <div class="line-out">&nbsp;&nbsp;2. Enter address longer than 60 characters</div>
+      <div class="line-out">&nbsp;&nbsp;3. Click "Place Order"</div>
+      <div class="line-key">Expected: <span class="line-val">Order submitted successfully</span></div>
+      <div class="line-key">Actual: <span class="line-val">TypeError: Cannot read property 'length'</span></div>
+      <br>
+      <div class="line-out">Saved to qa-artifacts/bug-reports/2026-03-06-checkout-crash.md</div>
+      <span class="cursor"></span>
+    </div>
+  </div>
+</section>
+
+<!-- Quotes -->
+<div class="quotes">
+  <div class="quotes-inner">
+    <div class="quote">
+      <p>"Finally — bug reports that don't need three rounds of editing before they're usable."</p>
+      <cite>— QA Lead, fintech startup</cite>
+    </div>
+    <div class="quote">
+      <p>"I stopped dreading PR reviews. It reads the diff, flags the risks, writes the checklist. I just review the output."</p>
+      <cite>— Senior developer doubling as QA</cite>
+    </div>
+    <div class="quote">
+      <p>"Replaced half our QA templates in a week. The team is actually using them consistently for the first time."</p>
+      <cite>— Engineering manager</cite>
+    </div>
+  </div>
+</div>
+
+<!-- How It Works -->
+<section id="how-it-works">
+  <div class="container">
+    <div class="section-label">How It Works</div>
+    <h2>Three steps. That's it.</h2>
+    <p class="section-sub">Install once. Open any project. It auto-detects your stack and gets to work.</p>
+    <div class="steps">
+      <div class="step">
+        <div class="step-num">01</div>
+        <h3>Install the plugin</h3>
+        <p>One command. Works with any project — React, Django, .NET, Go, whatever you're building.</p>
+      </div>
+      <div class="step">
+        <div class="step-num">02</div>
+        <h3>Open your project</h3>
+        <p>On session start, QA Toolkit scans your stack and saves context. Knows your frameworks, test tools, and CI setup automatically.</p>
+      </div>
+      <div class="step">
+        <div class="step-num">03</div>
+        <h3>Describe what you need</h3>
+        <p>Type a slash command in plain English. Get back a structured, formatted artifact saved to disk.</p>
+      </div>
+    </div>
+  </div>
+</section>
+
+<!-- Commands -->
+<section id="commands" style="background: var(--surface); border-top: 1px solid var(--border); border-bottom: 1px solid var(--border);">
+  <div class="container">
+    <div class="section-label">Commands</div>
+    <h2>15 slash commands. Every QA workflow covered.</h2>
+    <p class="section-sub">Grouped by where they fit in your workflow. Each one produces a formatted artifact saved to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/</code>.</p>
+
+    <div class="phase-group">
+      <div class="phase-label">Authoring</div>
+      <div class="command-grid">
+        <div class="command-card"><code>/qa-toolkit:test-cases</code><p>Test cases from requirements — table, Gherkin, or checklist format</p></div>
+        <div class="command-card"><code>/qa-toolkit:test-plan</code><p>Full test plan with scope, strategy, and coverage targets</p></div>
+        <div class="command-card"><code>/qa-toolkit:test-data</code><p>Synthetic test data — JSON, CSV, or SQL</p></div>
+        <div class="command-card"><code>/qa-toolkit:e2e-test</code><p>Playwright test scaffold with line-by-line comments</p></div>
+        <div class="command-card"><code>/qa-toolkit:api-test</code><p>API test suite — cURL, Postman collection, or Playwright</p></div>
+      </div>
+    </div>
+
+    <div class="phase-group">
+      <div class="phase-label">Review</div>
+      <div class="command-grid">
+        <div class="command-card"><code>/qa-toolkit:pr-review</code><p>Risk-flagged PR review with plain-English summary and QA checklist</p></div>
+        <div class="command-card"><code>/qa-toolkit:bug-report</code><p>Structured bug report from a casual description</p></div>
+        <div class="command-card"><code>/qa-toolkit:exploratory-testing</code><p>Exploratory test charters with session-based heuristics</p></div>
+      </div>
+    </div>
+
+    <div class="phase-group">
+      <div class="phase-label">Release</div>
+      <div class="command-grid">
+        <div class="command-card"><code>/qa-toolkit:regression-planner</code><p>Risk-based regression plan with time estimates</p></div>
+        <div class="command-card"><code>/qa-toolkit:release-readiness</code><p>Go/no-go assessment with quality gate scoring</p></div>
+        <div class="command-card"><code>/qa-toolkit:risk-prioritization</code><p>Ranked risk matrix across features and change areas</p></div>
+      </div>
+    </div>
+
+    <div class="phase-group">
+      <div class="phase-label">Analysis</div>
+      <div class="command-grid">
+        <div class="command-card"><code>/qa-toolkit:coverage-gap</code><p>Coverage gap analysis against requirements or test plan</p></div>
+        <div class="command-card"><code>/qa-toolkit:flaky-test-diagnosis</code><p>Root cause analysis for flaky tests with fix recommendations</p></div>
+        <div class="command-card"><code>/qa-toolkit:accessibility</code><p>WCAG 2.1 audit with plain-English manual test scripts</p></div>
+        <div class="command-card"><code>/qa-toolkit:setup</code><p>Read project docs, confirm detection, save preferences</p></div>
+      </div>
+    </div>
+  </div>
+</section>
+
+<!-- Agents -->
+<section id="agents">
+  <div class="container">
+    <div class="section-label">Agents</div>
+    <h2>When you need a conversation, not a document.</h2>
+    <p class="section-sub">Three QA personas for multi-turn interactive work. They remember context across the conversation and adapt as you share more.</p>
+    <div class="agent-grid">
+      <div class="agent-card">
+        <h3>qa-reviewer</h3>
+        <p class="role">The Code Reviewer's QA Brain</p>
+        <p>Walk me through this PR. What should I actually be testing? Translates code changes into testing impact in plain English.</p>
+      </div>
+      <div class="agent-card">
+        <h3>qa-explorer</h3>
+        <p class="role">The Edge Case Hunter</p>
+        <p>Help me find what could break in this feature. Generates exploratory test charters and surfaces edge cases you wouldn't think to test.</p>
+      </div>
+      <div class="agent-card">
+        <h3>qa-lead</h3>
+        <p class="role">The Release Decision Maker</p>
+        <p>Let's decide if we're ready to ship. Makes release decisions, plans regression scope, and produces executive summaries.</p>
+      </div>
+    </div>
+
+    <div class="v2-callout">
+      <h3>What's New in v2.0</h3>
+      <ul>
+        <li>State-aware skills — every command reads your detected project context</li>
+        <li>15 skills — 5 new in v2: test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis</li>
+        <li>Redesigned agents — explicit tool boundaries, typed return contracts, persistent memory</li>
+        <li>Multi-format output — artifacts in the format that actually fits the deliverable</li>
+        <li>Shared references — consistent behavior across all skills</li>
+      </ul>
+    </div>
+  </div>
+</section>
+
+<!-- Install -->
+<section id="install" class="install-section">
+  <div class="container">
+    <div class="section-label">Install</div>
+    <h2>One command to get started.</h2>
+    <p class="section-sub">Works with any project. No configuration required.</p>
+    <div class="code-block"><span class="comment"># From GitHub</span>
+<span class="cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
+
+<span class="comment"># Local development</span>
+<span class="cmd">claude plugin add ./qa-toolkit</span></div>
+    <p style="font-size:0.875rem;color:var(--text-muted)">On first session, auto-detection runs and saves your project context to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/.qa-config.json</code>. Everything is ready.</p>
+  </div>
+</section>
+
+<footer>
+  <p>MIT License &nbsp;·&nbsp; <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank">GitHub</a> &nbsp;·&nbsp; Built for Claude Code</p>
+</footer>
+
+<script>
+function copyInstall() {
+  const text = document.getElementById('install-cmd').textContent;
+  navigator.clipboard.writeText(text).then(() => {
+    const btn = document.querySelector('.copy-btn');
+    btn.textContent = 'copied!';
+    setTimeout(() => btn.textContent = 'copy', 2000);
+  });
+}
+</script>
+
+</body>
+</html>
diff --git a/docs/style.css b/docs/style.css
new file mode 100644
index 0000000..418085a
--- /dev/null
+++ b/docs/style.css
@@ -0,0 +1,159 @@
+/* Reset & base */
+*, *::before, *::after { box-sizing: border-box; margin: 0; padding: 0; }
+
+:root {
+  --bg: #0d0d0d;
+  --surface: #161616;
+  --border: #2a2a2a;
+  --amber: #f59e0b;
+  --amber-dim: #b45309;
+  --text: #e5e7eb;
+  --text-muted: #6b7280;
+  --code-bg: #1a1a1a;
+  --font-mono: 'JetBrains Mono', 'Fira Code', 'Cascadia Code', monospace;
+  --font-sans: -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;
+}
+
+html { scroll-behavior: smooth; }
+body { background: var(--bg); color: var(--text); font-family: var(--font-sans); line-height: 1.6; }
+
+/* Nav */
+nav {
+  position: fixed; top: 0; left: 0; right: 0; z-index: 100;
+  background: rgba(13,13,13,0.9); backdrop-filter: blur(8px);
+  border-bottom: 1px solid var(--border);
+  padding: 0.75rem 2rem;
+  display: flex; align-items: center; justify-content: space-between;
+}
+nav .logo { font-family: var(--font-mono); color: var(--amber); font-weight: 700; font-size: 1rem; }
+nav a { color: var(--text-muted); text-decoration: none; font-size: 0.875rem; margin-left: 2rem; transition: color 0.2s; }
+nav a:hover { color: var(--amber); }
+
+/* Hero */
+.hero {
+  min-height: 100vh; display: flex; flex-direction: column;
+  align-items: center; justify-content: center;
+  padding: 6rem 2rem 4rem; text-align: center;
+}
+.hero h1 { font-family: var(--font-mono); font-size: clamp(2rem, 5vw, 3.5rem); font-weight: 700; letter-spacing: -0.02em; line-height: 1.1; }
+.hero h1 span { color: var(--amber); }
+.hero .tagline { font-size: clamp(1rem, 2vw, 1.25rem); color: var(--text-muted); margin: 1.5rem 0 2.5rem; max-width: 560px; }
+
+/* Install block */
+.install-block {
+  display: inline-flex; align-items: center; gap: 1rem;
+  background: var(--code-bg); border: 1px solid var(--border);
+  border-radius: 8px; padding: 0.875rem 1.5rem;
+  font-family: var(--font-mono); font-size: 0.9rem;
+  margin-bottom: 3rem;
+}
+.install-block .prompt { color: var(--amber); }
+.copy-btn {
+  background: none; border: 1px solid var(--border); border-radius: 4px;
+  color: var(--text-muted); cursor: pointer; padding: 0.25rem 0.6rem;
+  font-size: 0.75rem; transition: all 0.2s;
+}
+.copy-btn:hover { border-color: var(--amber); color: var(--amber); }
+
+/* Terminal demo */
+.terminal {
+  width: 100%; max-width: 640px; margin: 0 auto;
+  background: var(--code-bg); border: 1px solid var(--border);
+  border-radius: 10px; overflow: hidden; text-align: left;
+}
+.terminal-bar {
+  background: var(--surface); padding: 0.75rem 1rem;
+  display: flex; align-items: center; gap: 0.5rem;
+  border-bottom: 1px solid var(--border);
+}
+.dot { width: 12px; height: 12px; border-radius: 50%; }
+.dot.red { background: #ff5f57; }
+.dot.yellow { background: #febc2e; }
+.dot.green { background: #28c840; }
+.terminal-title { font-size: 0.75rem; color: var(--text-muted); margin-left: auto; font-family: var(--font-mono); }
+.terminal-body { padding: 1.25rem 1.5rem; font-family: var(--font-mono); font-size: 0.825rem; line-height: 1.7; }
+.line-cmd { color: var(--amber); }
+.line-out { color: #6b7280; }
+.line-key { color: #34d399; }
+.line-val { color: var(--text); }
+.cursor { display: inline-block; width: 2px; height: 1em; background: var(--amber); animation: blink 1s step-end infinite; vertical-align: text-bottom; }
+@keyframes blink { 50% { opacity: 0; } }
+
+/* Typewriter */
+.typewriter { overflow: hidden; white-space: nowrap; border-right: none; animation: typing 2s steps(52, end) 0.5s both; }
+@keyframes typing { from { width: 0; } to { width: 100%; } }
+
+/* Quotes */
+.quotes { background: var(--surface); border-top: 1px solid var(--border); border-bottom: 1px solid var(--border); padding: 3rem 2rem; }
+.quotes-inner { max-width: 900px; margin: 0 auto; display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); gap: 2rem; }
+.quote { padding: 1.5rem; border: 1px solid var(--border); border-radius: 8px; }
+.quote p { color: var(--text); font-size: 0.9rem; line-height: 1.6; margin-bottom: 1rem; }
+.quote cite { font-size: 0.775rem; color: var(--text-muted); font-style: normal; }
+
+/* Sections */
+section { padding: 5rem 2rem; }
+.container { max-width: 960px; margin: 0 auto; }
+.section-label { font-family: var(--font-mono); font-size: 0.75rem; color: var(--amber); text-transform: uppercase; letter-spacing: 0.1em; margin-bottom: 0.75rem; }
+h2 { font-size: clamp(1.5rem, 3vw, 2rem); font-weight: 700; margin-bottom: 1rem; }
+.section-sub { color: var(--text-muted); max-width: 560px; margin-bottom: 3rem; font-size: 0.95rem; }
+
+/* How it works */
+.steps { display: grid; grid-template-columns: repeat(auto-fit, minmax(240px, 1fr)); gap: 1.5rem; }
+.step { padding: 1.75rem; border: 1px solid var(--border); border-radius: 8px; position: relative; }
+.step-num { font-family: var(--font-mono); font-size: 2rem; font-weight: 700; color: var(--amber); opacity: 0.3; margin-bottom: 0.75rem; }
+.step h3 { font-size: 1rem; margin-bottom: 0.5rem; }
+.step p { font-size: 0.875rem; color: var(--text-muted); }
+
+/* Commands */
+.phase-group { margin-bottom: 2.5rem; }
+.phase-label { font-family: var(--font-mono); font-size: 0.7rem; color: var(--amber); text-transform: uppercase; letter-spacing: 0.12em; margin-bottom: 1rem; padding-bottom: 0.5rem; border-bottom: 1px solid var(--border); }
+.command-grid { display: grid; grid-template-columns: repeat(auto-fill, minmax(280px, 1fr)); gap: 1rem; }
+.command-card {
+  padding: 1.25rem; border: 1px solid var(--border); border-radius: 8px;
+  transition: border-color 0.2s, background 0.2s; cursor: default;
+}
+.command-card:hover { border-color: var(--amber-dim); background: var(--surface); }
+.command-card code { font-family: var(--font-mono); font-size: 0.775rem; color: var(--amber); display: block; margin-bottom: 0.5rem; }
+.command-card p { font-size: 0.825rem; color: var(--text-muted); }
+
+/* Agents */
+.agent-grid { display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); gap: 1.5rem; }
+.agent-card { padding: 1.75rem; border: 1px solid var(--border); border-radius: 8px; }
+.agent-card h3 { font-family: var(--font-mono); color: var(--amber); margin-bottom: 0.5rem; font-size: 0.9rem; }
+.agent-card .role { font-size: 0.875rem; font-weight: 600; margin-bottom: 0.5rem; }
+.agent-card p { font-size: 0.825rem; color: var(--text-muted); }
+
+/* v2 callout */
+.v2-callout {
+  background: linear-gradient(135deg, rgba(245,158,11,0.08), rgba(245,158,11,0.02));
+  border: 1px solid rgba(245,158,11,0.3); border-radius: 12px;
+  padding: 2.5rem; margin-top: 2rem;
+}
+.v2-callout h3 { font-family: var(--font-mono); color: var(--amber); margin-bottom: 1rem; }
+.v2-callout ul { list-style: none; display: grid; grid-template-columns: repeat(auto-fit, minmax(200px, 1fr)); gap: 0.5rem; }
+.v2-callout li { font-size: 0.875rem; color: var(--text-muted); padding-left: 1rem; position: relative; }
+.v2-callout li::before { content: '→'; position: absolute; left: 0; color: var(--amber); }
+
+/* Install section */
+.install-section { background: var(--surface); border-top: 1px solid var(--border); }
+.code-block {
+  background: var(--code-bg); border: 1px solid var(--border); border-radius: 8px;
+  padding: 1.25rem 1.5rem; font-family: var(--font-mono); font-size: 0.85rem;
+  color: var(--text); margin-bottom: 1rem; white-space: pre;
+}
+.code-block .comment { color: var(--text-muted); }
+.code-block .cmd { color: var(--amber); }
+
+/* Footer */
+footer {
+  border-top: 1px solid var(--border); padding: 2rem;
+  text-align: center; color: var(--text-muted); font-size: 0.8rem;
+}
+footer a { color: var(--text-muted); text-decoration: none; }
+footer a:hover { color: var(--amber); }
+
+/* Responsive */
+@media (max-width: 640px) {
+  nav a { display: none; }
+  .install-block { flex-direction: column; gap: 0.5rem; }
+}

From 63ffd814e0bfdb5bf3b4e1a2d769f48d52beee16 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Fri, 6 Mar 2026 23:29:46 +0530
Subject: [PATCH 07/15] =?UTF-8?q?docs:=20fix=20AI=20slop=20=E2=80=94=20rem?=
 =?UTF-8?q?ove=20fake=20quotes,=20sharpen=20voice,=20make=20claims=20speci?=
 =?UTF-8?q?fic?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 CONTRIBUTING.md |  4 ++++
 README.md       | 20 +++++++-------------
 docs/index.html | 21 +++++++++------------
 docs/style.css  |  4 ++--
 4 files changed, 22 insertions(+), 27 deletions(-)

diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 316f0aa..3d2cd98 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -89,3 +89,7 @@ claude --debug --plugin-dir ./qa-toolkit
 ```
 
 The `--debug` flag shows hook execution and skill loading in the Claude Code output.
+
+---
+
+Something missing or wrong? Open an issue — this doc is meant to stay short, not stay incomplete.
diff --git a/README.md b/README.md
index f817fe3..f74ed91 100644
--- a/README.md
+++ b/README.md
@@ -18,12 +18,6 @@ claude plugin add github:cyberwalk3r/qa-toolkit
 
 <br>
 
-*"Finally — bug reports that don't need three rounds of editing."*
-
-*"I stopped dreading PR reviews. It just does them."*
-
-*"Replaced half my QA templates in a week."*
-
 </div>
 
 ---
@@ -34,7 +28,7 @@ QA work is repetitive in the worst way. Every bug report needs the same structur
 
 I wanted a tool that handles the formatting so you can focus on the judgment. Describe what you're testing in plain English — get back something you can actually use.
 
-So I built QA Toolkit. It auto-detects your stack, remembers your project context, and produces consistent QA artifacts every time. Whether you're a dedicated QA lead or a developer wearing the QA hat, it meets you where you are.
+So I built QA Toolkit. It auto-detects your stack, remembers your project context, and produces consistent QA artifacts every time. Whether you're a dedicated QA lead or a developer wearing the QA hat, it adapts to your stack and your tracker without asking you to configure anything.
 
 ---
 
@@ -97,13 +91,13 @@ You get back a fully structured bug report: severity classification, reproductio
 
 Three QA personas for multi-turn interactive work — use these when you want a conversation, not a document.
 
-| Agent | Use When |
-|-------|----------|
-| `qa-reviewer` | Walk me through this PR. What should I test? |
-| `qa-explorer` | Help me find edge cases in this new feature |
-| `qa-lead` | Let's decide if we're ready to release |
+| Agent | What It Does |
+|-------|-------------|
+| `qa-reviewer` | Reads PRs and translates code changes into testing impact — which user flows are at risk, what to test first, what's safe to skip |
+| `qa-explorer` | Generates edge cases and exploratory charters from feature descriptions — finds the scenarios your requirements didn't think to mention |
+| `qa-lead` | Makes release decisions, plans regression scope, and produces executive summaries — the QA voice in a release meeting |
 
-Agents remember context across the conversation and adapt as you share more.
+Agents remember context across the conversation. The more you share, the sharper the output.
 
 ### Skills vs Agents
 
diff --git a/docs/index.html b/docs/index.html
index b615018..fca3f42 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -59,20 +59,17 @@ <h1>QA that <span>thinks for itself.</span></h1>
   </div>
 </section>
 
-<!-- Quotes -->
+<!-- What you stop doing -->
 <div class="quotes">
   <div class="quotes-inner">
     <div class="quote">
-      <p>"Finally — bug reports that don't need three rounds of editing before they're usable."</p>
-      <cite>— QA Lead, fintech startup</cite>
+      <p><strong>Bug reports in seconds.</strong> Describe what broke. Get back severity classification, steps to reproduce, expected vs actual, and environment fields — formatted for your tracker, not a generic template.</p>
     </div>
     <div class="quote">
-      <p>"I stopped dreading PR reviews. It reads the diff, flags the risks, writes the checklist. I just review the output."</p>
-      <cite>— Senior developer doubling as QA</cite>
+      <p><strong>PR reviews that actually help.</strong> It reads the diff, maps which files touch which user flows, and writes a QA checklist grounded in what actually changed — not a checklist you've seen a hundred times.</p>
     </div>
     <div class="quote">
-      <p>"Replaced half our QA templates in a week. The team is actually using them consistently for the first time."</p>
-      <cite>— Engineering manager</cite>
+      <p><strong>Artifacts that fit your stack.</strong> It knows you're using Jest, not Pytest. It knows your CI runs on GitHub Actions. The output matches what your team already does, automatically.</p>
     </div>
   </div>
 </div>
@@ -82,7 +79,7 @@ <h1>QA that <span>thinks for itself.</span></h1>
   <div class="container">
     <div class="section-label">How It Works</div>
     <h2>Three steps. That's it.</h2>
-    <p class="section-sub">Install once. Open any project. It auto-detects your stack and gets to work.</p>
+    <p class="section-sub">Install once. Open any project. It scans your stack before you type your first command.</p>
     <div class="steps">
       <div class="step">
         <div class="step-num">01</div>
@@ -97,7 +94,7 @@ <h3>Open your project</h3>
       <div class="step">
         <div class="step-num">03</div>
         <h3>Describe what you need</h3>
-        <p>Type a slash command in plain English. Get back a structured, formatted artifact saved to disk.</p>
+        <p>Type a slash command in plain English. Because it already knows your stack, the output uses the right format, the right frameworks, the right test runner.</p>
       </div>
     </div>
   </div>
@@ -192,14 +189,14 @@ <h3>What's New in v2.0</h3>
 <section id="install" class="install-section">
   <div class="container">
     <div class="section-label">Install</div>
-    <h2>One command to get started.</h2>
-    <p class="section-sub">Works with any project. No configuration required.</p>
+    <h2>One command to install. Zero to configure.</h2>
+    <p class="section-sub">Works with any project. Configurable if you want — but you don't have to.</p>
     <div class="code-block"><span class="comment"># From GitHub</span>
 <span class="cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
 
 <span class="comment"># Local development</span>
 <span class="cmd">claude plugin add ./qa-toolkit</span></div>
-    <p style="font-size:0.875rem;color:var(--text-muted)">On first session, auto-detection runs and saves your project context to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/.qa-config.json</code>. Everything is ready.</p>
+    <p style="font-size:0.875rem;color:var(--text-muted)">On first session, auto-detection runs and saves your project context to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/.qa-config.json</code>. Every subsequent command reads from it.</p>
   </div>
 </section>
 
diff --git a/docs/style.css b/docs/style.css
index 418085a..46422ed 100644
--- a/docs/style.css
+++ b/docs/style.css
@@ -87,8 +87,8 @@ nav a:hover { color: var(--amber); }
 .quotes { background: var(--surface); border-top: 1px solid var(--border); border-bottom: 1px solid var(--border); padding: 3rem 2rem; }
 .quotes-inner { max-width: 900px; margin: 0 auto; display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); gap: 2rem; }
 .quote { padding: 1.5rem; border: 1px solid var(--border); border-radius: 8px; }
-.quote p { color: var(--text); font-size: 0.9rem; line-height: 1.6; margin-bottom: 1rem; }
-.quote cite { font-size: 0.775rem; color: var(--text-muted); font-style: normal; }
+.quote p { color: var(--text-muted); font-size: 0.9rem; line-height: 1.6; }
+.quote p strong { color: var(--text); display: block; margin-bottom: 0.5rem; font-size: 0.95rem; }
 
 /* Sections */
 section { padding: 5rem 2rem; }

From e90da6afe787aaa95d0711b1e62f7870e4bd74c8 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Sat, 7 Mar 2026 00:01:00 +0530
Subject: [PATCH 08/15] feat: add favicon, og:image and social meta tags to
 GitHub Pages site

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 docs/favicon.png  | Bin 0 -> 543 bytes
 docs/index.html   |  17 +++++++++++++++++
 docs/og-image.png | Bin 0 -> 40711 bytes
 3 files changed, 17 insertions(+)
 create mode 100644 docs/favicon.png
 create mode 100644 docs/og-image.png

diff --git a/docs/favicon.png b/docs/favicon.png
new file mode 100644
index 0000000000000000000000000000000000000000..cdf577cadcac71e39267fbeb1d4f54d8f06de5c0
GIT binary patch
literal 543
zcmV+)0^t3LP)<h;3K|Lk000e1NJLTq001BW001Be1^@s6b9#F80005hNkl<Zc%1B)
zPbh<79LK-U>s?#z9S0X7k{rZ=IIM8xLY9;e<+77W9MmpWTpcGlVzqJ^Cn>jy9JGTH
z2Q_8ppwO0$?S1^-p_a(r{du3uK8N@Byw9J0-}m=>o?q`HBMYn5I>i`sP}5Gf7Da?}
z9wBo|HV}`;x4-+D_OMtijj}9nlGA<|0EswErBcpBA`#9Au-R-uGCeR55)q8WVto}9
z_8Ize7Yv4cHkeGNTbUYT5QOGiCffmVbTG<4hd^WqrYxW=K)3)Y_o*C#ITa&aDvC+Y
z0){&|?9IUXG~jclQBzf%@=;+nn?1Qqyoxg1KXG(70d4gRlXi~RH{d+1jWXnm?n+o`
z;aGNXOppMo<p^E_qh1}E{DJP3FA0W4zk~(<=hC*i!cw986lf9fPz1&)0xIptt>gT|
zuUo*RU4?rHs8)c>2!m^f;o{C90R!zEYvVxjonv*M!F$B$U>4?FL)RvB532B;FoYiq
zcbu1iQc!Ry?rl-AGV=S_*$uF?E0tD&Vg_3OxIsjp{|~P46iD=c{s2K}zO@8}AdZkk
z^?U<@(0peOw4x{n^fCYgq44j~)yyG%DVe8^B#a}LPN!!xzO#EpT+crM00960=v7Yx
h00006Nkl<Z0{~#VnQBR^G>HHJ002ovPDHLkV1hMW?|=XR

literal 0
HcmV?d00001

diff --git a/docs/index.html b/docs/index.html
index fca3f42..997b844 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -5,6 +5,23 @@
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <title>QA Toolkit — QA that thinks for itself</title>
   <meta name="description" content="A Claude Code plugin that auto-detects your stack and produces structured QA artifacts from plain English descriptions.">
+
+  <!-- Favicon -->
+  <link rel="icon" type="image/png" href="favicon.png">
+
+  <!-- Open Graph -->
+  <meta property="og:title" content="QA Toolkit — QA that thinks for itself">
+  <meta property="og:description" content="15 slash commands. Auto-detects your stack. Produces structured QA artifacts from plain English descriptions.">
+  <meta property="og:image" content="https://cyberwalk3r.github.io/qa-toolkit/og-image.png">
+  <meta property="og:url" content="https://cyberwalk3r.github.io/qa-toolkit/">
+  <meta property="og:type" content="website">
+
+  <!-- Twitter Card -->
+  <meta name="twitter:card" content="summary_large_image">
+  <meta name="twitter:title" content="QA Toolkit — QA that thinks for itself">
+  <meta name="twitter:description" content="15 slash commands. Auto-detects your stack. Produces structured QA artifacts from plain English descriptions.">
+  <meta name="twitter:image" content="https://cyberwalk3r.github.io/qa-toolkit/og-image.png">
+
   <link rel="stylesheet" href="style.css">
 </head>
 <body>
diff --git a/docs/og-image.png b/docs/og-image.png
new file mode 100644
index 0000000000000000000000000000000000000000..9f352e2f11ccf14dfb5db45a9bb8a73457490248
GIT binary patch
literal 40711
zcmeFYWmr_-7e9&xiUKMiB_JwYN_VKVbhk=(H^YdON=XaD0Ma?c&;yKsv~&(B%@8xx
z&<y{>_dfso{=U4=^YhJ-12bpueb!!UePVypP*WfwrX?mIARtjve5XY~aJzzl;HLN8
zo4}pl&$sUr5IiJMdMB&no4!4FCy;!dv1^aj%cp$~_(w=5^i=2raa6;r;MPvC)LL(6
zdMov-r{M^ML$H?7CQr}RWO&#JOVQBHRVkPKP0j|y%M^r)Z_6<vU|?Ww&h!TG2!iS^
zwg0XH-(>Dy|4b|Q?Dq9F`uW4_A$*p9bocuAGC6lcuWx5*-hOj^9b&u5bbYm@2!42d
z75M(YkN<D!{uegDR`}nR`Cq!YPKp0ZmH!{g1d7daZ*S;O|2&t|N`p-Tj|u<Pk0Y(+
z(ghasALdQiZg}sFGko-TqlLSEfk3_)P5XBn7h=a_P_B2nARL+XC37)F;jjy~!EnY?
z9zgy?bwq!6vd6e-w_xoF7w{df!Y>}q?|S-JIpoZn!#ax5Nw$Ix<G6EEhmuWZE;aO0
zAqU~cK<H}DtPhWg?ykN~8wK#G&j{k+-Nwfm;pdxlLeR0DX4g8M%Alu^`KI{d)9<Q{
zE!{Tk0`kWb(vu5$)v@!+&$vGSGM`R7*0yOc1AXr?#aT=jDj_!~81lYf7uQ*q+ihDc
zUV>AoO({bVRbCuNaPR1<nW6$=o;Y1bpMVvJ<OYLS)wApIia)x`ujseiY8_5*GZwyl
zSal1yE5ms=l!gKQ!L@1<)Y$nxxOb)lnA6)yBSK!Pn#-s&YbdO=+#mdP^Xl7l;HJNS
zb<Jyu+0LP+s$yDOw&k4c$7A0#<{~)9lC|%AOW{49P~ZO41bHo;A|*f_AQ`Q~rxVmF
zH1j88Ae>d4Mr=}x@{&Y9+lthSv3vyDl6DY6Enh3}!0}U|*&F8kP14(REk-3XzEwjb
z*kyW>a-3lLec?m=5Q_xn3?aSn_R8~^ZZ!RYuG4%kKE|npW=+cZ6==$wQSfJ@?+(7j
zdBN1<gy(vu>dzlqziVOYe-Oco)??_E>YW$LQRoj}_3C=#>EK<|X!QUr;VhE$!Z~lt
zYk3-N3iU(`6xVd|HI6!J_6%8n!33)x-suj7sBK35vVEr#Qdfc*9aqkziEFcfBg-K<
zi{<yijq=7~_`@@6H+oN~2?c~aOPn?i;xO+vRI%s74=+Rdnc0f(1bD(?inUc6Xc#hc
z%qt2W@-6N@Ih#By=F|)if;;ERR^rBRG4_sL;jChiSZ^wrX=c7Q$=kA#M&z;OgV$ni
za@Tzlo7U|&g89*{qjJZyj2!FZEa5e;6lrI{vd@!MmHj)X)+lk(G5jm(`@!l?UtT1l
z8)KY`H?9J;nuoM;(7b(zb~otkiW=4*JrCR5VHQ(?_DYOOz91Qx`^AEHZ*~`HtO;6^
z!OgW7&&?1vFP_00$M_U!C7xvG22!P@zphZTYtL3XBMYbY&g3`wTA#;Z5-WD*LrC*{
z$?F*W$%FYUKP8@cr89xXCw-gwVH!5+(BuR+;CL}N>X4K?n0=n2W(7>*^Nie9!C%oC
zI{CWf{Vb#E!S|#zQ!ih1239Qj(^AX~OqGgQTKE2oGVrpfxBRtn`)z-`oG-gK9aMI|
z*v#c;?QZ99Mj_N~bAFwBI;xzrzFQG9*0WcxVSM?rZoApeds0w%i1UY0<FxkL@dxIy
zLVsRbJ(Y22Cf5X>P2j^+M${HDT}(hWm*U_h8W~h86z*@OW7PO@{G%?x@VDN%hP(Xn
z@1)3V=H~BnNMQ1l>dzEAAwn?M%U>&<$w~ZL3stv-`=j>W3UZ`iWo%!WT+P%sEp}C&
zC3*_(@<Vc%JUg%V5`}b{l%>w|g%>-x^|o<yME^@+u#c5Z$^GaSx9vR27TB5Fj&2ui
z_+v9MY2gqx8<vmCl1}#02RF9soHn;`r+qqBf&6NMO;3|9O?l+aGW|viTq{ORqhIId
zE*zevGOLDUB6@k3^Ozi+#)K^nwJg%leB_%o{|l0E0n;g(5w7KFpG=q#I6eck=I2+0
zZa}KhFt&H<k9|$HUkHmDkc$Yi;@>!kdXNkf3t=*K!RHTN#X&khSXb=6d&|JucXdvi
z84uGZ2T$bX(`0Pg7t$8$U$5YIrX9Vzc@ig<Htplob1pOeRQ1o7`SJb@D|js|>BxEg
zCShQ)kX6Hvw_5f`JwoXf=nEJJm%arbevNDIj=$vvNpu7L7p%kT0_UZj%A}cM#rJm2
zviP@9Ms?~1VsUga+)tghvDxPCw|uqhv_`={UG`9FNHOVn6jU?ti7~^ir|}0{2qb~W
zc;VF%s{BqNZPMTgVpWHUHM{yRjYox`wD&qH>VVE6NnmPbULc^K!&+k2Z|$so%yX{w
zLkP4CX4e79n9p|aT$o$``xh<+|7T;dI7+P&(}_Ir-kdj1c6+>b!HUDE)s0E3+FRF2
z9@SnWsNYUGC>dMd>SnX=jtedoLc?bkG+=WKX!%FCknUmgrj@#sK3|_3u6LGtq{zCH
za#MGWl1fJwO48>Gd)=@~ky&c&@y@Stqq7dV_k`YodNXMp+G0QDh3QP}<bRpDom)3-
znpss{_faMxN}}QSY)^uIHa~uXvt)m%c~uC!qepqm6DghVQ^9=_A?YEw>@rIH6m&hX
zSt&7}<BMZXt&2H*^D7)RZeS!zMcE*=)ctc%2b@q}VXj95hkdlV`e{Sk9NJjCnLLcU
zdgp}4W*@X0Z!838$T3p~p|VFv8ux>g-z=xTGED-K03ojT;$}W3pOWN0%SH~l7t%?4
z;HcP|oifvs`Un^O+(-`Go?Y-D$l3lt$?GJZW8GoHL?th`D9(IK$Xe7_SNSVgeN336
zT-=S>LYbFpyv4th8C$@sl?i*z??Tc?ll*>y?t%TN(p<;asK_U3riJ|UB*rC)ANvo&
z$0JDEuOQ?-sRtPX)up8qPm@vb*YrQ9b7ER#8=}NJ?~f|{Daoue0Y@5^c;-8Npgygq
z#e>YSSf<g9CG8c{#6VaO+_O^Y>9_0Up`bKYP*BK|gomS<KS`b@sA?70G(CD$Zro^o
zT3@7$4{x;cqJp6_E|m1yWo(+B2JV7WJf&SVeJ;NL(o-6*wC>zXuxT5cxta>BQDfcf
zoKxbS*Z-~P!LC)e7YdVC8voV!%Zl=wynUbDTns-=>ZpjU<rhrs4vm~>YNq!%iQ19<
zsw{40Q0V-N%XdHj5KJt6V}*gP=+IDY8pmSZQMx|wz<B?Rl~hsqyRCsR_S7o-A24h7
zw~Ezb{)Rr=y*Oc#U5y+g_R}cvP7B^MWaJWVzLaUNrAf6?q{q-AwMq^>w3tDs>Bd_^
zUjISK_JEdm5G0K!kXs)=6=Sf?jqzs1rh;#~XvJ8?+t;)kK~$cv|AX0C-L{o)rfSnZ
zVwN+kr`i867Z7e(GG=^5k?YO+(OpA3spidp$h{-(NdOHf>z9<6ENG9{jTBhe&-r$*
zDI{?arve|x=j0+!9{oS}4#h&LUMPhCKM+)_<v@F9zu$kf`R?m<fGibMvG^MPYed-{
zR*_%-0YNka-}O5pc=rDrK4A8S8BEA|KnNBi1XB|t^%lEB&>^dw#w4Z0TUY-Qq|;<Z
z2^RD5@wo#u9|)1@Esm6mx;l8YO0ANj<~{aCPs!kiGJ78!EL`1wK-@jX3i9(ZiIr5<
z1Phm<u1OB(Oak?psvAd(l|qwCPLu=`HbKtl7U#1AeeL$_QQb<bt&NRBV8Om;<hE)B
zky61>9V|0!1Z-=MwH%k?9eYovH*n!6D@#VKE9=xiZXk_05-TFfKw3l6J8xtYU>@y<
zut+=XFSAIy75ipu(-6>V!S%`(+Z85C7M`)OZJbnE%=kE7VIbv8OG^}=v-KN7K?bF&
z_F{Gqe@n^*O|)4!t*~Nt!zgqgxREjul=-fq*1Mv?0yDpNcPnf%g6z+0%ZOz(-94&`
z4BF6+BI_hK>xFf9Riw$|DN=MU&gfTvXcD9i!9Wq|b{Z$Ar-Qx@)|DcVMy<Tw8iz?u
znZ1>@Qp6K=C*WzE&&^n+J&UQAerV46#i*&R_9gLZR#|FjeEXf1_4_jc_=-UK>H!!p
zdb~`wpkJ>)+v;_>5~`FepTuKoy&Lj&?cyY)QCxA?VWLUyLlzgqqmAV)ItuEQo#huf
z-aE6Rfm0vuP`uS*ic;1ha4lC>(q>T8a^wt}i++69Z5_1NG0AOQBP8BC5&ry0a#gL&
z!Pt0)R?yxkHA}AyQMf;mW5dh2N6D5y3w45ZYBJq+9aL5tZ3v+33+D`e_`7*<uDyL=
zV0sXmQkQa5Bk}2u(*uCT`j?U9Jk&!5<p{qUYOPbE*K7y3UhUIo(}xA7R#}C7Q8u|x
zCJkN;+zq?GJ^-DobHQfJA;!3=MQmO3lyyth`44u`#4=3ZzDGSEAt8D2pjx0cb+mvI
z*o;RH@4AdsYOe#YQpja^WNj^HQaPD}x$X(K5iKKwgzvve^B{acYhcK)XH;2%_WS%T
zEkP3nsAh-tVqH#(&u^+$GUdo5B?h(&6B84q*YUrAAghj$@ipz!D>F~sH_B5^I^c7I
zig_($oDTd^O0WT4EPg?*6<yd4SLwp+G(0z^_Aq3jtGy|_y2YN~9y8y(6>@QLk(<~U
zG&e1cKT&0h%zDgh>GNe<<44{Kmhs_3>W2(Zge=YkbW3M%ZJmv(OJU{;Tx4Wq1~ieE
zXML&sIsiBL`3Znj1RN$V;Darm-flgU@LV?uQ0Gi-t<V8J9dc~rwLQhon^5DhAUju<
z-+ZcfPD<Y|RH_aYEBb+MWoqw?2LBlxlwQX@0N!}f_iq7qrmQbtN=R6*h+|vKbF8^X
zWtz>dws@sMyLfTp-=?MDAK+sGWOxRRzS=xFJ>tUt<bv~~AIO<A!kL7gPOsN?cUu%%
z`$>KOqTjFCz1mdV=RDt8NEf=&f`6yA8B<@4t0X4CaNmA2UZJ@>*m5F<v>WQ4nDE}L
zrg3M*rdD`u)%vBrFaAnGCpZlcsB1h=<klZ>d?TY592~5rrQI<rSqoFokZ=Pdr6m22
zI7}O(A|krROMXR=F*b@Ltq7u&Ma7zoOYFtXT}FRosKrR$x;#sH5Qy3RrqVN!ZOU#w
zCr;K-SF-H7+#ppe2@#PR;#)n%TJV~}y+<Mj76wi$We{hue5P9N?H%L#f|>)xi|1nx
zlk*Awnf=1a7}Pm8^CHhC`ukNW6JL|(=jU@@oR^!h!nE@2!@g7M&U<2Jlkzz<Hj=qb
zvk553Gsny3|Mc;%yMFtcIZ+gJ1-qq}agMfNP%WN@_&L)@s59VvKgd@nt&2(Pw!y+Z
z1{NPbveIeeYz5Yc(g@i-b2aa0PJCx`!qU18?Z=+$xi$ZG)o@G5ubBBJfZb?K_=X&5
z1?)F>*vSn7o#_<dcDrXDC8)mJ$(_?ahp!-n(Uyttv#!UEmYFK@pyI?e0%>Sa`rLBk
zN3bgG_!jsI#J`Z?cJ!A<g<M=zG@AanuvfKY!pWWV{rmS(QD3`7o9xH!S2=Yw0JeDq
z6dZIXBTJoM!73@~-nLN#XNfx$NVShtiQ3|FIt^%(e*V}6G9+9I?>dN?tE{X9-pbj>
zBcB`vlKUhiHHV4kj2qaUogA)%y*;O)td9eW5oGk!b>K`?f#=S&>J5S{rJC6}VhPzY
z_uAew4{r4j#!M-%QnzjcCDHcyX`4Te4v6*HktxQY6a3Jx5EP*@Z6Gd*3W@iwMHSQ|
z5M+$6_Y({m+3z<VZ`gUCw15IWS$q=n-1zCCGgC|7g5OjLyYlt-^Foi*s6qYlczi}i
zhJHKDtkL)NNn2Pnov=wuAUn2vG~22~KAJX#P4f+3Q%cGt4k;z|nkM_jix<3vM43{c
z&40m{Ta)G5_2k43rVFxs>tv)f6ZuZ!U}-R_ab$G#OGpT>d86k__hNa=g@mA>?i`3i
zANCxn)1Owii`(<^^6>CVl6H1;bKBe7b91YVdlW}RMlVtUUZ^cI2A%@2A8hz)y1N0#
z7z?|^>cCtETko-Ok}aQ|Sya6eIox8VX6pm<!&;}w`R!e(xX(KUh;&V}+2z>?MXkV*
zknK>#unY1k^%&hHz5jETD;ji~w|<qDiR=?f(<#(&+vgwj-kF};Hu&b28MHr~&Iqp9
zCkFNx2(vZW0_vLf+7uBLO@(&8&_<6~ZH^T=Jbiv2yc~{W1Ru(}?%v_#Z2_&D8ESL5
zSe}b|w=FGpf%~rxR#VJ0>gJlOVPO0!nTolVilvV_LWp~x&uY0-H3&ccS9klxb6AaI
z(4{K#JISlIzDo@FcWW!tmcV$)<PYO6HRwzYc6WO;VIWhwoLF}>zrO_{*)>0Z;ip|i
z1Kv9=NX`)V-Tg6a!5|bs(QM3eN?4@aTm4}Yk67zsv~ihfl)E9f=IUq?puo6AZt&Dg
zsu^VJAonKq_ekL(yhn(nRPcx8%{v6g#5)Xp(^;}FRC10mC=2G0o!Of9kwRKQ2cyl3
zF%&A#&x60ee`UqePIG4lp^(-f+NInMa}Yl0Us~GOM!}0^Sfnqvzl2>GDBtDa?{2>6
zbyxsh<>cM-$6??>%M&)40kUcE<dH1H#|Qq$4cKxdm5PVq+!HDu6OE1qdXq_?@3FBi
z@|L}&(!54Zjzj;5;CZd3ma*+k$R|oc+F=Uz#a98xw^Zz0&ySpvM?N&4k*xKbi{Scc
z{IItSfg*yAU<7A~y986rP-+OQnLJ?My)AtSDJ-qxlkNS^W~B=wCy;raJuJgk7RXL+
zNQ?03^5z_zgFB?;Y1Y=(S|g%0X`%~atUbDia}-{Z_d+kvkJ3QdI@M#%ODB~%))90M
zp70VfM;=D<BK=l|yAI-)EEi*`MSHgkxgo*@1qCHv+<kmf*7n;ix=ONr@E4-G7jvy%
zremdnp=UTcQhK4KEteKv!{Wr=Xh)rLb&xh?hH!q1VfMRw@SL_bn6m81He1mH@WGBF
zCo*DUyv!Jzvar^IJG!FKxUCl|S9-tB*LDAU$)x4YNQ>iKnD~c1jxbJd*6Wm(kNv7X
zN9SVaxEUjoDs$s^^P8Mt4ev!`nt+QxZk!fm3{~FaFW^=(R-%(TGXauB82H1xkUHCs
zb}8UJciiT9Xy|Qtj;9zj&xV9fN{jdr#~=e}`yO0wy1V<UDZ;zh_+a*cw-&3ha(WT9
zI^Y-?DUN_erX|Jhj;+|31%T}5krhcCx(rqg4kdwT=?@=b8JhpbjBMd}C9WI|3=9S|
z$BT8ZmLkhtE=R=o_5@-yb<D(r#89)CrjzPw{TOH!es&W4O+{@O$o=s_z3L5))k+KI
z<B$LtlV;{&zb+7+52Fd&SX9%DXb?VF22P8d=P&xrnA;S+>mc;=#a<5ZcAuT5AcD;!
zC*yv#f18^VFjFGk3)2TY-Eluo#hSEF&d$6x$2ztP1xxgcjaj&ZbQ5!<?22mjqGtEJ
zKm{vXKU8w0j&a&`rJp*K%isI;k4#~a=hl5|s@4jU!0gTD=Qk(P#eI&NeTp7t7pAKz
z7N1~~zY)c9laL_e6S;&NUXYc*=4j*7<NG|`9wv3lzdz%!Mh~oTK0>=Ei0@ZSpM3ps
zw;8ui%ahkB>C3UW{9~-F5<RK4%Hp%~al1l0+`$egZA`Q~I{9zzMDEo72@TySG#MbN
zr}ZA6>AJAPSoJ0Mi&pqWCXe{`i<%f+&D5G=<7Kr9n`+>ZDg!-Yp4;_}uiE}Xz4n&m
zyo_~VS{>WF9g$C(FTSVBduVBCDJm+uyT_N8mm|TStBU<`v}Hzhs3yA{4%!r+x%@nx
zT+B&hi;AVAQ3`KLxkjt9rl!=Fs2X=_dzhn>)56uR1r?P(90(<eU3>ucCp~jO`E;#Q
zITujAB)zx0SNl?+Q0T2&H~SK~jT=2v(-WHU(-g~YSefBeF!OuoVwp7;wKxdJ51`F)
z*9I!|sO_`E-HItD%lD2L>B~L0CkVj=>(uJODkyBCNZSTZ&M4{r5`#>MFW8=`o7}D}
z1%|8Z)1As+km-97=XfhK(EU?%TFqZI<Vt3TN($b>do0X&OJi@SHRNFZie2gNzyQpq
zd|qAcr=^{;ik&$z!nn>cnfWOLL#@wX#@h%|2GKx+?{ug}GWx6T9E;AV8j2HLYx646
zTZ;#SYPE<{nO5gH|JljOz`ga8Jqw_64{aBk_{dX0fld&-+bV9&OKeo2)~vu7U_#l<
zU!XpihFhGjxnf~uzC%a|Y+Lc<0iNVWTO6(6N`VaJ*a2^oNshcKJSRIlJ0n%1M{9a}
ze(oa7Rc9@zQrHaawb*tC6`sOnSnPy2nrl3^9nQ{k>sI#&s_^s#XhF6o0<?D}o4E+v
zrlzQqAX6d^zh1w3#Wv$LVq~o7d44qK=~zs0ub;3jD=RBJnih+4s^o`czc{t*-_%7B
z#!!n+j~r_K>lM8$JKu{yxgfoA&ZE_(PYpP{pa@Sv@yew3qZa&GA8G-N_}K1mg_<-6
zH3Xun8o+ofv_@m`c=77fp9k(?H4etxc`t-mYT1%UxEoU@Z~(^WC-w1cqMwYPuV_q?
zp83CAz{j~!byb0#tvIn6@vDG<UtuzAn`1EJe=@dt9`rrI<*I(CZFeFUo8EK~C(A26
z8%!k(yQsRv&+xjJm^Ac`*^PoOFQk00Bmpq|oj*F`t!>_$pFe-9D2zYv*z~(UFHH2!
zeYMZnz82FWy1Y{{izv#Y&3W?oa{0LQUaKvzQ<C!mFj1YSsl`;CK^UD&A_+RvQL(tZ
zyrldt&ji~Nw?b7hFLy~%_#DXI)*`QUf8~mDU)2d#q&(b7#ZFKCyrQ|fJT}*dnK-Ya
ztG0Sa3dgEV)Z)avHdiWsg>*&H#{K-61qdH*`5iKaqv+DZ;*_-b^Hf5#zXR2)Ztqef
z?EG;m^nLlHOOt5&_H3o(@SBCGvYFY+9C>W1(@bs4(Pp1$L_~z4MCwZ<DB)*~{kXEB
zthGX_Bt1nPMy+h?O=5EL(FI%`LhmIgFg5}&P5)>+cQp^<efImpu)HgV9(p?AxuqT(
z6;&-@-~HpGMVDl@JNaXGLjd`FQi}|J#a0-a^u(Z1DkY;URMw@8=&NH_PHnAg;nX+s
zE_l)-pcf-!5VQ4OvL7#2$<wqO48A=7OB!1lXo)Q|aT$6k%FaIHhGa1_Gb5J!cRik$
z;*Yen`^`~Y?CeUE#Ssh)Hbp8wLqkt`5?EHIha;1sIF!y&>t+Tm?)@Qe!vRhR*y2l~
zp@df41E}d0hIpLb+Z^albl&FjvT>am2%fOBV3U7}0AOmqDkK%!;xLI>2)xXDli#*8
zuF=cY5)@QbRi&4{y^RU$lb*kYiDux^9@WxaEYdAG$}PyK74SZTXb)!eU)m?n)Vb7*
zVdFGU_bKOpo~`i@N+Io*!?}%xE)mxKH4cjaPc&;>O?|f=W`xKsEi=XR{TY~_7BEDl
zqvvx9%F4=6J>C$6+SvfTcqh;nqR=z74(iXixV*L|5>^d!<oow2IdqD33~zKqz~OxX
zSg>VLNV?qpE4%|<B8lkg0)Eo4Zr)q7w1dIS1K7Ty`L!UZD`uxJg~zbU(k*>8Bl!k_
z%wBef%+HJr21ds9%}s0XQV_%Dn~8!Re@$W~dh3EX+U4+QYky%>r3GZ{?O!3>=A_F%
zG@6E<zRdML07X4Fh8CHXlXYbF4to<Pxp3N)w6tPQn+_Z<@d@!Qmun?Jcf}Mt+SgJB
zp!VLKuoN@17DseB@*=RijR^yYzSP6Eyg8MXPkda&#l=fz4Kh1DdlR<1UFMs@P<8_u
zlH^t}1tO)82=402?VTNj=T5$c25wW>pf(T(tfxBdAJc(euT7HPe`^?PyVA@|^9J`-
z-3+B?CY7I3ds(TdCW+3-E<ErHT7|i>%)p)%2g<cbAJt>eV^{qKH=pAQzN1-`N2mCX
zxM#)K{Z9ZV5DOKtIGU)?v`{IZZ2>8S-L#u*Y4=&UvR!=~=I2xAWqGsgbYaGY`Q;x7
zJ+oXBnty&kzb-F<MPd(C04H|M&Tmq|N*!~XRh1gmYyz*C!q}+nVW+`%-B3R_S@4j%
zZBrmR9^GaVma$+|?a1!+CM>WRh@Db(DIX6HZJ9l)$L>J*Au9!H<@=422j14M0QiW+
zHwz1J6O>t;I*g19vC^l1x7H%dZvuPjT~6>T5s|48wj#6=xq>0(KgJJsW2Xn}?~je^
z>~e^cN=hJk36DQ^h+{z?iik*a0DRwa(&o_~{$O*s%mY_U8C?gr8BBvtY4OU+e5}-3
z8ZC;_Gza540KSG>^<X0>HicbVZ?&r@fz`}^?=)s-rb%)ch~BH{=;;1)kG_=h-IWA#
zfUjZ8sGcf{nyPatPr^?)OqABxjS{L@s3dc(ZZCbZ+x|w*8s(EF=%PtsGwHcK<p&f1
zV!w%NgnRk5^UoiNv+ni)Z>*loUR+w5*m8f7Iq0AnRk?plaHpTfLOxdGl;V-w4uBi_
zR55jSqcAT>ZNG7Kci*F5&F6<u{F=_4mGG_loAr~FeOvr?0uM2>m2<sQ-;+3u?uAa2
zYgKsf&@MyOppAvkhl;(J^y#sFbBN<=KJPa2*my$Cr)B0y%z|Kvxc8=`w_{{-*%=wN
z2zWlTI7`Uc1o;}yx&i>%C|bdMpmOW6B<lrjO*oZZINp@k)YJsJBhx$+9k<-!xAFkb
z@rZ%J+RhOwhAWk*1Mpq4d{}t6^GsuhFVbuHWgSok2}$UR#Cx+QjcL887}N|A-t;Ot
zqzn=jpmU)lb-!($&R$C)q0^0n<pB<}Eh7iKut#_j9G?L?3VS1}`0=WxSzSFtWTiE@
zKs~dH&k)ULb&Wm4fim3DdW(oF+I~JeEQwXfWqxF&5Vti!i%sM-Zvo=E*w(X3gS_6U
z)FeUd*)xqftJ0^xJP*4cC4v1OywOXJ2g`isVEI1TIFWX+K6v%&JV3>w85C&d@AsYK
zWx(-v(CR8x*?WyfPrK{N$FxJ(MnX4XOuLF@eC0ay0iKgPI|krf74-{%Jm?f}_QzEU
zM4Oen%$LR$tv^<&qwc%Rs<oR@o6~&#dN^5L7p5f#jNou^5#wB0H)Y+QreCNbJNIS7
zHd3YNeQMSGyi~CW8!-KlmXoE?AAKp@(K~bNEfZxk93|ng)OP4~9*8!px&e)YRR|HF
zf${_6SqQp3U6P9gMN93i<nFKS|8{VB4?j$FL7r~@O&#q#cbYbE3<cP}k!@x)998EM
zn|Iy)rx52hdQ=_drm05L3M$0VLOtqNGP@YS`*<S4!kK|;E%$gw7a{WA;>^9Ht3<1b
z!57W=7QNW=fE}*`N!LThSGz?$gT)B>J)=q5#Pr0&BjeGs_q(0d!lqn4U%VDTm-$U+
zosiK5Jm#EnYrLeq)PA9L+RL+)Q3FT~7$cyq?o=f9AED+G=K#3sf=uxFM*Fd9GoOIN
z{8C;@flj$u`}U#DV7ec6ZtGE7S63ZuHGdorI@e*?o`V-S%^W3ndZR~F_X`@|dShno
zthW2gn<jp}E!df<Q<^_IJmfHM2}~*;rxmc>*w|ou`{Y7Uea_5gAl+`Nuo9)IMVxVp
zJT?nD4cp+ld>RE_83x+MwvLYRUWZzT1*XY0EdKHwXbidNE2e=8f=W*2HTOr4bR2qu
z@*X5@0kvQ!K~vOZe=s{iujF-|DzFMHP%#L0q{v=3lHOWuY6b$*v$XtUI=?g2o75-l
zY~X_a9XyiP_uhXqwQj+mf1M2#wE*Dc<x9P?nQm1xh1#C+61_F9W3R3G<>;b``1Q&h
zh3n)JE<k@5jJF$Tw_s$~%nn6W#y_!c!Xc^5)1Ipyef~_qNyQ5A#~{m#_WXzu_bT@^
z6bve^!(f2j&IKA*pH~ktbMW)o1~|~1Nn}^a_363yiJDC3=g3FD_O=Jlq>H%y?KS8j
z;G`9tN4U6juc|OBCkn{%ZROsywy;Rp;xIruJ1qnq8`B#9`XuK^56{U+7qJ@&WgRok
z@-1_C@hvA4l!0}LDD{C4PS<#>?|$?X%ibrAfBtK`(%sedmALoG*SN?hJR*T78WX{)
z5TWyb?XhrG_989keMFrPHp5|Y@#Er0wU`)LS!VmOQmae*8{$5g`oPn8z*gGv!<&po
z&>xqSQtCNcz^Bfipd6DVz=jy<iVa}flFKeL5jQKJEMKTlTWtk8-_mq>HaTi_b#<@x
zzvmt>jvo<SlAl~$%AIqVuP!VusZDuZu%z^20P@d%+ok5G6vE+AvH@%_6IO`ixp7Eq
z*1kH&?{SAQ4WEcl)iA*=VEag49p>a!R)Ou$qUzuo?~Tj+qt&I(N5;qdN3>U{@;y>u
zZ^IuTu7YnoJ}E0`t*#c}0Yo;Ly*!*uz1?UydwEk~-E3Wp7pvM+Mk&wTJqzKOr1&SU
zkx!l=n3BAW&D5R=h*c~swdU>fjF@Qq3}BI$pH?5d=ggfD3;e_8MQ4h5f}e;I)(tj5
zTX>9i&hj>RZ*L4Iai)}9B<wFIsH<uHH`JPKfOBv0#}(%Rb&O6LuQ>|P<q(P`hP-m*
z+4C)zXFwpCRWsb}kdCAR=y3!`=F7MEqpiVP=f{2CF#Uo7)AaOoE=kT<)`zq^bBM0q
zsiU1TuU(C?m+X?q+oMHAMI2Y15&&OE74wAqEC48B9GSJwb0ay%tSW{Z3t;^c3E!P@
z{Wovk=-|D!$4+_JA3rJgDl@!y?;f_9f3){@tkluFR6eWdWALfpz*`IX@bK_r<AjV1
zDT}i)Lq;$lgUM9NJ^Q=6BWw-=2(*b<`hIzP5kR#Gx{S6XJ7oN~y5mIa-29>i>L(sN
z)a@A^05__&_p*UZU#IAc;XN&|_NnLY?;ys^8PM>Zd)6iTWq;2PrTb4m{w-Ym`0?Z5
z`wR-2M{!Q2)wu^#!-}3dV;h?7K=sMmM|Ur_!}P>!8=%)kfD5%(X>-h@VgWTZ$!`{I
zKXe5chg%Hw*t&%vrT2;&cb=asv#9xA6o^W50P<(LU88ntsT6)A1i3aC@Tu)j&?Q6n
zvUYD0CvP3lfa#<bTwERuE}4T*(T7FadOVcP!}}d%9@EVMDM=|>srrOSPjq&p2T*LU
zWROzoYQCE-EtgM$CR3fuF91M9(u1e3jn|*lMTDlZX8KlptiyO)E|dYTv-Ev~%fftJ
zzPh;U2jt!&kgXW%uUhKn@#YgS?OMnA!u{=w1`jQzpo?$!UdTuF_KAM4io3v{m)~Kt
zdr|~Mca7ub*yQ?E&CtNW1sTGz{gUW`04|;DTs$A(o(teo41)ge0sdGek*jE*MdYN-
zXCdg|+y%)7O!@*{Y63Ba6i*foBd>nPGpn4Qo!&Vi>RvE`AX(y{n}K(dauj0NqT@Qo
zv3Me*I_Cu3O|tk_Af3X(!!yLZ_PgU=Y)#<lXlQ<6Zj#Un73-G=0eH)PqS&;|2v^ws
zfQc!^XM2jrid>uGZ@weS{aAVk31I)E5*_I2(V>+}O5p{az((tUDMH9@<Sz<^5_vZ-
zCML$QG+hI<bQvN+j?0}SdOYdA!aymd_0uoW+tDjC?(3wXrdZkB++14Pg!*lV=yY3r
zviO7Iw;^Q|tH1h<2kFIH6LHy@^`GuqdB43?pEAyoTD#+$&Eb4Ev;mL-aAbFPakw(G
zW_gH^6SJCusEWq)ajc|R3?U$Z8v^$C*@L^GMog!>r&Q)u?v_P~%JL5ac6k~qM1Vql
z>n79Z0JGO5p`IKGE{`5fng!xftc>8Xk=p>G(=Jf`&yeHl1Mfe-=iYr?x8vgCcDPJJ
zTHf`<4-ab+9btjLkSQ7oI{H|-9Zu-Vd*5}^yeT>VJ6%xPeN0N`x&hkIa`vR_+Koss
z+i^FvZl#Q7^pk}}e+=>6yBqU4`JZml{$DPj;6KMcf--nFIW#N`B)0WJ#lpDOvF!S@
z|D5H(=5_W1>5TGGv{0PUI;(mHPy=>G@~)l&1kJTA>G#}s8Q+ozgzaQ-WWZ_C<V$>+
z@iDe4Pem<LD(}DVfGIrVx-ESrVAFnvqvA>bZ*m%e5AWV5vb#P7AXEAF*|mFsAUZn*
z@N#8w-@u&7TWM1ge7@GrtJfAMdbItSi<Z{L&Azb~DA2jIr!m{m8Kb0k5{Q<e+lP8)
zAK!H4?(lZ#?5k*p<yzpz<X4Gz#tCL0UArN&nC5xTZjflVpf(D>r^{xiL=R*U_drzA
zOWyA#8h`qpT1k0vIOkA+GIBzRqG)F6`&s1!j6>cjCoLE@^8UIPtWoqh!yg5s(2%%3
z)9^X`fkX1+)GTBFZZpbC9zm@XX-cf4xoz<7L^EQfVU5nLpuOLdY46i<E*T*U+e&6h
zNKm^+W%RvHw%Y5n5Lwi=+RR=9KkMT&(3?-Drm4e)R5Re?vh2_{)AkwB%P`D01~m%Q
zXYtQw@&GJ-?aK1Evs0cMN`1U&-^Exb_JnJg#P+9K_3ErCsUeZt_RpyjwX{qK{<qwW
z#jjc~l{~WZfqs!<m*$lWFD<&nalneRE;{3*F9MCN=*O9+=__8M7q`Z1LFFC23q!W!
zXQt|-xA)C>af<d@Jmalqb;n<tN|#f)bP+eAwFm{?!!+N3g=VdLWFx-N&z;F<hT=)I
z6$ZUgAf}urVnr593H0TRc4Dh81=8MgzWt>+YHG2`$g8In5Q#~_t&+KgXR9clZ;UXv
zjJg@78M+1$YAO213ytnZwo(1PFTAnk?n0!D6I`cZ&PrYderZx;hb{NU%@6pbJgbBy
zP*O!6{ng5N8YWls3e`q4{oa{}<k{|*Chci?8KiLZdFqUc<l`9~CC)7i>fO98eB>n6
zzqi~8{^P1;8wJvysG~?+jIr7bWUT-kl9+<`_G@wYbjgP%;z2dT+t9Y%xEi}0eu+`$
zB2Z0re8yQwiX(Vx^8{PDrGDr>femD~U{=Govy1htReJZ}R0^EL9gc@$Y3O<??KEqJ
zXdp}L?T?e<A)EWZW*QtCTwea33n69#YAYd+9E$X6K$UhcL5%qdXKUiOa5IsZWO@4K
zRb%^W_8s|ig)hCKG3<!JnU?*!gJI6vw`sGUK^dXxcLL}$&e~5ybYo;0&W;bfHCAik
zN4ojGQg{`4fnw=?Q;dyj$Wr9Tz5d>UAVkmJWl{_fy$ezAUqS7s8r?T%j#<sVR)EmF
zHa*`Tw_B3xN~v!D6e!f}eIaeC9<o&4Oqyej=;jvWK4&Uryn?vdXX~iOnoZCQYW__o
zlpCC8zrkSdy1bH(70lP|J#`Qo?DsGWn8gfxvy%yQkIuG?O8-*n7x2^TI-Wa9uWNWx
z5U@)ptStdwiteDvY+F#h`gHuxm`>1Ht&{b;Qe%tHhYvX`J*M%;CKU-aG?}F>Pm_Om
ziL-B&fNS)Z_S=tKc-(kalkkgKbPfmT0i_&cmU1ck<IOkgW1hI4kUh!^nDlFoVfo}9
z%z;piij{=v#I*Vhy~X2lWX?Mi!wz=^y*f8^9HHj62GgIpfyqgi?fSiFTHzv-2B6v8
zZD|fbUSYNX9LEW`C=}BZM*0J8kw1~9IX@LE;GGnw-8bRbHWVKR+|YeT5PKl1Sjr`>
z&-WL2+`*@dQS-7lXI<X0eDlsfr+VQOM4j+&C!HbHSA`p&zpA~JGE8>S>t*gL&+mA^
zj_?}nJStA8I+2kd$Hdk4K-8wCq@^u_k{R4xy<SB*LDS(pHXYq-O0jk7{!98P+&0dC
z16^~=0MBmgfbGSCdqMrm3JTIc#*sIBhUtykKx`J;o4wu%4rhNXRfbFptz@VfWjIBx
z{>zJEeuZOP4^#i6Po^yG&0KzeqTNu*&TYj}n0)~?gU-HE@g^5zxoVnQ1o!c;DJ0Lw
z->a;WFgN=5>FL%dm!G>370WlBn9tRoxW#{e>}Ani<x^r)wBdp(rz_?Kyuu<qg<#U@
zt8>ro_Okc_F3+B`{1*^bX^AecEt>OMT`h|Q(G}`r(wj^ed&vMC(2fa9N|s{{%Jj+_
zXU~ohS0Tp>sRzqs!~|KA`kx$gbwrBgMiR|>j0dM^EOgBj{aoi6h%6XhRn@#6<BG#$
zWNk)&jCq<)4CxH>?(7}0<N9)w_D5z=)|Lb=`jn}K&T@UqcK6tO1!humoS5x}*fU#B
zKEE@5o3L-k2=nGsZMsaqyd)t%h<`(p>M>U?yV)n1V0O}H)ejJ*-6l|CIFN=e+({#w
zA4S(@*6=eVIzYMJp9U;W<5-o0HSuel55NCy4=+!_M!n@Y%}d%9aFp<~kebY_aL`4|
zoAcWFL+y63A}Oy^OF{QSqkHojMA=cvc!);bR{Dd#W?!V-#=RY^JCix`Zeo~jw@JA9
z_#m%>@kyNeN#f4V&M+A4+qZAS!^1#@zixMc_MO(>>Z%1a2@B(-$oT#Hw{X7N@$PQ7
z&}>f+8XzgB8?E{Pf{6lSS|y(4w#<UTFgRl{84L_Xc6OSYal@L!adG?o2yf=YUbOa7
z?#(Po)!w`q=hVlK$`P~)lW$b}l#7yXjWC|C=7%uN(US-XHB#TdG1#soZS(%3ry{ET
zF0d0zO!;<>_k>CLo{qdc?O;IhS{t~e74$|&Brwg0ce0>!J=uL`pOVxHHE#T&W8~mC
zAOV?>_`Ux%a71%wG}O{d;SckcM7l>NmjKI)q@?5VOyWLYgHG0EIOI!S=<c1^E}MW}
z3uOJz0YU*WU-kh1!(Nkc>6iv|#XNj*t?uD6$B#QaU`Z%HUY>dOC!bhP=cs7FGYQ(Z
zRK0rPM|V9m5#CThE60o#!Qm>HfRHyl=JQon2Y})+a3ebG({{ard%o<8fPiyWmiK}7
zO9{kbvb6`lvz{fXf7CB$5UDVy6hb|EX5N_5eAW#$VU79r?b-?j^sCJdlWKQlZl!ON
z4Em_hP6(W4hLF18mtWq=mXO*5HZ4%JkdHk^5AU(l(ER>M><(-|?jdcG^MODGzLJ<}
zl^(^_%v_d92wAhI58pfM{6{o-lXSB#6nB|dEH9!JySKV7ydpq`nTKSA%n$I>Yl_`F
z&TD?+TQA7nV>hZKd@}1mFk^S$rf3~y1p##_dTR@gA*G(fUySg~KfX)K7?s~YH}bi?
zCFp}JnqTJs%w)v=^?AkMKnw9B&Mv((YM{;vnlK=pow*x2@jmV1uSKk6+%Xzx$Wh?m
zG7UA3^Y4_YiR<#^qmA^sCUn?dlEu<509-|qI;S2NM1YCW{6`be_pNW0s~;Tg^#CVB
z`mzAj>khdX3k6zeHiF}7`PGs6oy=#cuCwYeI<>XsH(!!2Mu3K4OSF%BW;fB(RIwVS
z6n2QMGhtlHT)q8<6%lp*t!Oet_TC`G-NH9LdN$Fjvl?w{^A&meOqtzDX{AJ$Z)bNk
z%G$uL{BUIc%qi8aJIfQuI0D?6_M3*)hNolXzl7q1iQ2|Itz0bjEPo3>5!b&r1QzD_
zg|%_kw>U_9xtx#`(%f<P<^A0~Qg*no_|wMpX0&all}w~E`kcoE+?6O`waLzEQ3bK_
zJ$qch*!Ab?m6()jIp8`6ZxB%dhaC{Cd7Ivuyo$GQp7qdDRVsy~vSi!m2RVHr)k4eB
zFL(MDnp24|7^E$ysEIWmscvIyA_Z#iX&QK|l{WC<mB{p*998=5{V*BZjDnQO6m!r_
zmh9UkFd$d5kUSwKQzzrJkd$nuz;KVw2RZ<@2wl9KoLn4XaDdsd5j&jW{YgOiWi|7F
zThFPuFZNpu`$&s$fb0>Db(>H~nFiQ`fuX7?v_+3Q-`gDTh`FSsS^pekM=uMiIhEwA
zViMgFKYm8w-}`0|hn#<Lda*p7K(DRG*zYS_Btk>}?~MFN-{LWnn)a%nr65-a=em{X
zG3l60-<a`n4eRxDg!2?iv^1Gf;bJ6J=5jekMP(^1kN#(cAfnCBAnyAa9T_Lo>CEkw
z>ZgoXtfIY2=#wU23Zf>z`n?V_MCjCfx>P#vnl5<Yoc{Ew-Lhr*5z&lUBy#T$RU@lc
zVTRcNE9kdA29+UF<>#&w*!c~5i#AV(bV%LY9poGvJ|#B2MVqWT<IxI}O|9N@auJKd
z6{#r&@J<CGnO_`tGNbp+Zq_(Y&~{Dwv!YOnBk&~Tpr?|*Qyd)qoq|u=oFw+j4`ET0
z<!P}l*%{+)>0#vYH|SH>JAI<do|0NJ0mZl)$66wK_z13%yI?0=NF{03xmzODV=Faq
zyNm!Ma`P8WOiYY<(5Y(cNY=@6>;eUGI|^=7tXF1jRoW)GvZ`;!X+TS%f;Y@1nBAIy
zRxlofY^{x+I@=TZZ3K)d)8L^cs7frC&%V#7`W7S+CVWvj`Vqwx@i=3WH#SIAET?X;
zOZ=sazm)u^kw4NVL;~+=P|ukiCUf-u6imH8YhOk4DBYrC&Ln{wy8UUTzMID-x)d;3
zhfz=5cuoeiT|JzTGi4}ze7Hkaa+>E9A8yxE6dpSv!reV=x2^&fRo_W?hU%0r$jHQ|
zABYiE_6TWtK#pxMfVPv53^}HXg}ka)vR_l2I}B2Cp8?IUwYoUxinI5snr;i^>nba$
zH;=uPHeRnjI*I;Kd-&sR|Ko3wN08Hh(#~GnX5wYeXaQqrTEi@=`b+Ng(UDeHEMZ5=
zZai()PBl0kN$tJ-_qVyXh}684-BF=0<Zt8&bADmLTH{ZiOPjDgu~2uLgZq_ynaxrS
z>{$ad%NuCNpF^)v<F9kpPO)z1KQ`zx-^^Q`i=GRgvm_t#J#BD2*lwa*7jJpB;HS)s
z^F6nOs!=^Df$Snm&>|m4z3$Q_jgr|?NTPjg3~5no_Pq=Hhq;5{ETNqnmf#z~4?7~S
z51%huV_S|tW}crF-)sm(Yju%2Q4qgHc<&(3)&Qn{!S`}}0emj^>s8>%FsskekKtA_
zQmB|yBS#^?2d0a97~QCk6f9p0JlTh$NVp$m`)l}5>{@h)Uy>{Pp;f8!yPPzpIYC<M
zD{AA~5ado=yqS&-hQuNH;!Ry{qu&OOh^@kid9;=uS>yZ_dz&K^1$YULw#SVr$1YC6
z92D`mtw})C^a4C#9x-B`yB~h#%A%OGe!b~<%&bg!=guk7?c0Qeetv$@bmH=hN3;h+
z?SQo6GOb8^`$Na@!x>$L!l_xg#z|$g0XOax37d4}l}EuuEGNr@iW8cDZ2*VoRqlqO
zNEl!DPE0H{+-Pm33d%TG4h|vWAvC{x|9%43;_~wH`Z_>%pIY}D3pkFQ94`dvvPtkZ
zpUjq&t6qx-rMG;r)2G|~3FF*_ricLq77P%4|1TG?g2iIjI;H-Jm#ad&0snG^fRLEj
z*<b_GWwrvpzuLEdhf6!ZG`76FuFeOL7#+ro=gI<NrGwU=IT1`;RC1QHhAbY{Q&^26
zoHsG|Y1TQ5qcSXjoRt8Szi6eUM$+H(znOAS{m{BBfWom>fJ%Xi^<!ar8JuDHKYsKP
zhm`@Gv~G#s!NEa-qbq*40r-ER@o1A=A7ju3>OTu}V@(O?#X3$npM*wicDt}m%<?cV
z#hhMnI#5@Lo!#!kRgfN!2%zj1^`-E#6ZkZX`*z#5=p3ngRO@$q(u=HVjZR?oQa<t#
zewkm$zbJIY06!0cTLhgG_om2@f#i^$1DLmQOFxDK-T_``uIMw%358yckq?hhjaKkk
z3XCS8jaCWR<C_}vMzWrf++;H;XtSlABh{*H?oXN-qJ|7nbNu*BbFn%1r=z1|LHhT|
zl*jr22SHfB^0Ox2y=;JNk%?n^I1jj&7kZ~G6k@M`)xRISr4I`Q=rjTf8s+uVe;eI~
z=-*Cx>XV%R-Qo!K`HVoEe+~{Q(Jz1WkeZtM;l?@^`|saF1_p0IfkXSL@yOOzaDS%$
z*xZJ=m}vA<(MGkoMrM^n4%7T7INU==N5|J-D(v5Z!(^#plkczA4;uAWH*VaR3pjKf
zZ&U?lI%l$)2`MPB0jx+SCXSBpJCWpFv5X=WF6XsQ*QCuI+)(>+6h!+o__Lb_aY7kd
zxiE1uhEZC`2ZJ5Hmj4I{x`SnWjmHZOJ^p;vG)3UL{c1u0jibBY*L?a|+}rWWVLlVk
zLDkgN>-N6@;ynZ~118;DK?C+YAWsXCRMynci`xG!945PS=Z+PW@0E7hbWOnF25@{}
zU}jdoYjjXT7$4WABZWhc&-ur`kT(aPgJx=XbmtY*#JrcH@+tv=2tC{k?+pPwu<Ai)
z4V|iKD^FJwdwc`k0V$4%l$Of}VD}EYut;TJZNIghoflhfyZ?v)fv+l>1BmPEGn7)t
zlKUIWoXNZp{`H9EE|E1FU}nO@p3k;XB|dYFI5;Wbey!xY<R5GS#7l|KPHQuxNxFJx
z6N8<do%wzTPuvsM@Z{<e?##LW3N)22013DQ(0o+MkodI#b^}8{Wp{TzNnxYMfqQRY
zql&t_tL)`o*m>>Mh<AH-qYM5QJ>N1xXM33*SKVG`H*OI%AMR7=PtKO-umvyzax7aN
zz~L_eFiY1bal35;Ria)S%)pS5oSz~yll2i`?x+B1OF8TMb==7Vywk<HTliaqtDv*E
z9C>3}N<fumeyAK7IKi4(=QwV(nh+Y=0Os-C-%}ucI#o_w{4uAq^V%^Zb#|h?Wv5}%
z@{YE8<CALz?p|PEAQB>3pqx}(ER*YazIIrLMIx_4Ss9Vu+_NK=%Zz?Hy|Gnqbb}O0
zYUQKT0Ckp{j;;jrLpiA=is3+Gz5qK?14*j9PldlG1zqoN=;XVah`8+Be)a9+9mNoN
zE$6dM2o{KsJmO-rX8%rc3kF(;W<RHQ$QV77`X4G`yFjX9bEOri^l`07BxJs=mihVs
zZIdzR0{F;BYv+K*%(!BG-U*0bHK;_<f6urZ8n)cD${@?&?`yIp<TTE`dvp|Z+)Sn;
zJ6@_g0|2v`8P-@<v}sGAVe1!wD>k4x->^_0D>PJ@OUup8O)>Kw7?}c8Q1{=z0~2dJ
z?8oYRJO+`4-;pXQJbC%~0Oe0Gds8fl{ptE3lwZ6_uWXu}RaH+fEX*`IGP*>gGCRpG
z_tDh}Il0u;c7kJjS^B8$bGr(+gnUW=!~OkzEkD=2PZeM8*r*wAU5@w4c08Ao!u$@t
z#gxLEmzz&`Cy7nBI1y#EBy0}$+e|!d(+4z@t{6I(Rh+QPJRolbS5#CORjrN}U3a8#
z?OwI-pn&7F0WR+RFCq8md9PDP)b$c<(;S5597~-{_4M?Nz=JMV3(ArBktv%f1R|g-
z25O!u)(>c%+kYN&3A;?!gRg)Z0r<cr_D)`Op2r^p`l2*^5cDtj2H<<s@Ox2S<ctUd
z^L8H%M*&t8X&_L%9Dw<aN(WtWZ~Y*q<d8a_y`P*f%srA`%9TC<MVy_S7|}Y^zNJ!E
zA6(-$0(9E4w_P!vn2&(qvJ03+xSG4~^R_tJ9f-J}?A{-gw%GK*VecHQVgVmf^7m{M
zdg14hwHsMz@=+Twqi5LbDgF7jEJ_4aUc{R);^d5dY})u}ev7j|5ON6D#V-z%Fwog)
zN;2O{#o`NiUOu4t0=`ljYznvWz(@dY6WVOv1<*qMnZlE!qg#uWC{DGB!3eJBfR`ZN
zcwkYqd{C3%{0P${^zPHoyl>iCTJIwT%sKRnKG}^ezK!_njb$%!>Dae&KRDd?$23p@
z-*`D)!fWgw2iPM#F(c#Kg<roqyij=XEi$rC=wdlmYIk?{M?cHD_De*s)J;DJT}SAP
z>vM6le82$&eX}4<XTu1^e4_^cvOM(Gcu`z>dg=aq4r#zN5fLGY|CwrWCFC@{yFR<1
z5cSmQ$Ph57X@`BB9Du4z>|JYDpkrL0=n<e25TEL2zf)4l(HeCZ9ao*`NYppGu$+ka
z_3Kx1K00aHo`huqeC#wJV0*Ys29J0^?`&bQ=D3KF^q<Ge)c-5@bMF&!9Z&P>EMZv_
zuxlgd)T3A)12TzgN{)7FHXxlYz(7EkHCb+KV>ec1rNR}xzkh_8VprFftNXN<9rhwz
zn(f}7?J0-JHg3852CP<MDmj2rYIgv^Xds$3)~!rS0jT2UIR9pR>;kyvu=GCFfXh_H
z{B|MW>K8YX%D7IoMl#R1I7|^<b)JnW291Y|fZct2-YL&Gh35$=?cW^j0ySmk-Vd(~
z(w?pGO{TLwTW-ioo}H{)sG=-kRZH8(V$1i-W!vR}DlH_wxsBqVDCevPFPOpGt<FnL
zOKO2t{~Z44Zlv<gv=>)&yhN?l0ZX5{ZdVLLjli|*wm2(3At4hmfSzfVGclQXWPh%8
zL^*<P0<qhb`Ln|9;0PssA>cjxf|65D%g?$9so@5=N33)m9i#FzWto+)OSmmHw3mX?
z?h?qilr(mE@~odeWphXGS6oyz7*%}0?=5$gs(YYEn8R+~BW<a0Yd;h66B&eGi}3+R
z4OEt1r3NjIt>y<F{foiDvQHH5+-W`;p#TQC+H1J&#0I!l0NF*R{cfmS$IMLENMU!k
zqLr>^b%)FbI<>&z-rCDYo&#iXIEhL1hD!rGMLaL^q*OIX3}19j()}Oa-Ycr<we8+@
z=~5OpmRRW`Akw9GQ0cuBI#L4CdksZJM3fEzsSzRcB3-&l54{swlxheaLV%EDKfeF(
z{l?z=U?1)S$M8f)M#}TM?>VnIpNHBREn%eD695scO8pBafK+8sr!hkS0AKb$`*Wae
z(=iH}8I~FT!}j1oijm%(;1Ip^G@B8uFE^#$ZQ&UqSILdAZ3!Mi?@{gkTT%n3hX?ZO
z6w4Z8zM7jR)NdMo^h`tk_w`NgUpy?!8i(4qB8LBHG~wZAdzsHhlUHR<im8Fs&rJa-
z4dV3cu2AA4!e^qKk=r%0HM{kHKP1<ft~eG;pXQk+cxe^G)wd%&5JU17ra|Xdy-B^V
z8p2Ap?c`L0=kDh(OR-Ml45H3iZqHpcW4O=v^{M@N1kx*mFVl_0I~^|p%vp`<pPL&i
z`u+O_j%n<t`@eK_djOTLbzNMET+cDB-SxoTy*JQ}MxTJiuwU%)nb}wF^MZmP_4Vif
zDHKxbLPfu3-fq13tiOj!dCi~9y3Kv%E`LJid@-c`DIlRT8eXKQU4QO8S+VfdD;aS4
z8E-9`9lR(%lEUxE9~`m+V(8oo%d0O%7FQZf#=M5hpS?ix=yAz%-BKVDw%~J)h#_vH
zlEV?3j;QNQ+kh5>p4M;8Vs8Dz=b=4W0)g}jtF0~f$eWe=GkJLIjEwL|2W%eie;9rp
zA@Jv0rv+x#&)zG&HX3>J3ky6(C9WsMCH~n>sme^U8U04RiFZ#AeKqnJ85s@rntZWK
z%{v2|=saALpG&!>F&LN2Xv(wBbI<9oqKpg^)igB?34a5X4N}l@L}WRT^4q4pb!+AK
zZ<Bi0MlewDpE{xway$Oy6CZ+K6Y4D<SLS<X78tw$y_)|T4R^T;xIg`0sY>Qsc}m`|
zxov@q%fdwY0>AJQJmb^xzq(^_eA(~7sBuYNLh*Jwc%qupuWoLZy^n8L40G)49hIz{
zeSla3bABQ<u-(M|{AQh<Hvo9Y8@n}9m5kH_vumx0`hEw99^G;`eKRR%2HyJeKQoN7
zK+CZ=Gw1bI=c<}?aQM3_kVye`$Y|mb?0!6Jy<&6{H-g_xF00xb`JhbmP6ZL<ht}KL
zj2{MKSGhpY@sC4EuW+CD8OTc2h?ID*awq2mORet$IOlLjegxpYisx4}Kyt#+J|9_i
zlT-U>Ptd392prO_vIk61)4B~&IpP^kkS5l$&0A~OsugaePgP;n`W9KWK7{BOvp=au
zranB{-3_F4#FzVO^8iACFf~UpZw@n{{B^-73?vz*#)^`V>R{CqTHkH*!xc9V0B+X|
z0g^{H$GeG_&$+LiK6jZWdhcAlN<Up?w=U`;E;k|{PU*N4lbM-$R%l${YU-DEa@Oww
zVmL**S4WFAZ<w_@uhtB={@mVn_usUOm&gA8E#Wvs14s?SwdhDDX}=&+Y_ahveY!Lu
z-c><HR+fP>?Sf28>#BI6ShL1wYv&=Q?L9rjB-yVUpbmmg14ngs9v)jLEC@aFoAbk<
z1J56yH9sV*g#7$Tu6LWuWHQRN?LziL_{@$80LJ|M`2mk%kx8-sDu^ushU#fIGsw6B
zyLu(W7|b^9XB*}_1BMR`!ghc0_ML?fk?hc89BKTP17cMzv&}o9MZD5Jr-t>ca}90_
z^=>W)YkULR)`i*&My1oFJz^!;jb<AH;+?+QPXNreerII#u#?NcO8!eY1*qzVi&^sF
zi>Mi{%C^xat>#vkzlOV}jz*<nqj%+EC{TrtMm{E*i+Sh=fZT$2<>^$3(Jt!o%ejtR
ztDtFOrsM>F;TYbDHPNfROlL7@d3x;rV1_r@>H9CK&|FJIU!!GtN<`%RPRbc#pE!Tg
zO#~u8Le6iaUqD_y@OeA8tyFdeAYs9Xp{}K6)bQcW-v)$PkfPg~t2c~%N+#%NswONn
zZ}?+Ja>v1*w=q%9z`)QDPFW0~hzq;OhgZ8ISsfi63oPB7sx+j(4-ST&eHHAE>MmdE
zT1QycIKc6eTt-cYU@h@HJ;eUJa&cY=%<C%t3vHsIv#&269FWT_8dx9@hzciXr0u&W
zCT~_C<UAToIKEDtS8Eq6cPn@9_;V3tiT~vSMES+EpS_#BYf#fp@dv;q`cf(ngj_S+
zt{?RHAbiJxLFmOd64iThJ3=Oy9GT`aU7WY?v{8$dq$<eIpD53U^<<VpDr#^2W8~1p
zEa_Z})}xV^ao=WXhZ02(2?WY1823{iCHj3TV?wjtZ?Nl$d;8{%qOygCwl=?Bwc*Lt
z+bPh*8#f$u?t}v5i+l>JLRwxG3IPtlN@~Kt;ot!(ZQlIt(-S~9yC(NBM*%qb3nsMH
zFsIzKwZveZKefBw1M}1!ed%a>VSStWO||eO7Y#)z2u6H#E<mtQOtjUSF5&C8?m(f2
zpS$*Tzy6@s8Olac{53DNZt5)a>D?ed-CmLh1Fimq<<6W(TM~t7CH(yzKJ!pKK0rq+
zcnp>PF!D2ukTYJ;KG|r{{JZFpUvmh4b-lVqf4jFrt<SLJG01EVQd~w|46Ll`U89@$
ztOBv{99bBjRjudVQY6R>;n1aoPv2=!D#BX#xhkBNPaP02l{?M1_Yv8f=dN9E8)<>|
zhg|$&A>UhcL*e#&&hs{-N&5O}6$`vZelcgH3o8X>N&0YmjmRIu;ber4b_hzHBAc4R
zYveq$n<zg(9uiU9?`{CtHpdtQK$NJ_T**-K38zE+3piq@!J);8^u>sk^$S?*j2|YX
zr%SnY<0@M}?>kJD#jnVFo12@%9hs#4Q#ZD^=BeG3EG#U*epQj4;L>z1K(c|D$+08H
zBzhpQ-0>#dA3L9|<@9An(r<s|bYQz6Jma7P^C4A6!rnxW>WB1?S^@L-8YO`nqj}RW
z|BA56-cQ2L#|DKD@wHK)wJsPrK|wDt^sq2XTfGmvq>u|A*9!q>{ydLB>le>G{l2Ru
zpALgb{Q`0VT^x}=04zpc5VvkUM3r(=l;eXWT<)h(RzNb{kQH6;h6BNzS`mNn->IkZ
zlRA@cPL|W1{Y)_8fw<1rvI^W}k@q(|iax>-UZR8rJUqC#%_MjB#H78Cbvu+}ebT%4
zk+$dNIb5@tu8K*G-^hJCgPZR1Dzg5*6F#kTfxqe}$4TZLW*@(qShj-&<ys$C-O)DN
z1i6Q-PQ>Vs9+1)%23VOZNp_xP!_~U(O?jotY~+_?D2z63tM6pH9vMoK3w(q>M!Hl|
z95Sey-@j@bf6KllK_^@Ka+2FA5g{L-9X6+>V2_yn$*Qm_p9XZSJ{oKmi({tpLU=eO
zE?Wvxsb^3)@#+Gx{rGIWtni$Kl%!UMNMl&UTPP~!TB|p7uAUDOCd9+VRb_+|V+lPg
z_R5%anJOLd!2-CvY`X~LLpTw&nrI<TovdSQA<ML5BfUX?YPr2D2x+X`v~0Q8;@7`6
zESC+>3F<3ylRml^psQY|?(XgQJK-s(4kJXVAOJzMzX&C<@I6?{z=lmhdHjCdMTAdn
z7M<}Qt@2Etwf?LGgXh4p=bn5&q})spx6F`Pdy5;9^|%;B_|}JVvc-4e!5*m_G+qtG
z$u<YL&3_47%I~Yr5_go$D*QG#+<+nJR)zs4mWm25A9~;<5OIkHJ6Z!-Rfh}wZJkhe
zhKE?&lRg4OFeuq!5HvWrdhC3$JsG16hV+a0|Ne5LHJFg7laZBjD@4#F<H?6R<Z|7=
zf$Y$0+2rfB*^xJ#BU^fzT}a4#Kc5<;7WE&w%9ABv9dG@(DX(9@?v8_X9_mgFefo>A
zgK5`Us0Ns2JS?yl?<ALwAH`K_sWpUfby*;hNHqrAgEjC?My(^ll0GJU+V4;gZ*0H-
zc+)xbbxO)Dr#NulR2?fxGdaWY*@6SPPcK4Y1{WSeG3fq<0_7G-$ad5`rG9=m<C4<p
zYVX&^Og2o$-S!tzKA9QYd9#>%jvhGT_KPSf=*`rv=+hT^v&er6mEqoP&2N-3lvqva
zaM7h$OifPS@0qXF<?V$U4VO?=Kc&&6GZGs``f)zkYJ>p^=&CtDwbRbFS_k%+1iZcx
zh>!D|*ZZDkdV~<Rrpe^LDqH3FaJUw7cP&U-0ugIYco8H&T=KfL7#W4MyD%+IcFD?k
ztIVWPYX5cZtraqpPDHh}_ke;O4U^pQwlw8~&O3yhode*kWAXsCh{+N(@ZO)c4i{X7
z-(5d@+zg;Te+r&!q{T_$<R5Ppxww#j8*e+&-vT6`WZ>_Y$SRe1#-OA0!j;s6m9qzy
zn|3|1m@i=q0}YOyor>)suskK;>Gpu*d%<9aM)NA5n8Dz}j)gLr|J+&!#A_<Y?nz;d
zcbi?*_3RuNxeM^}6heDLJd-RpRIM+{l=I{9#CVdE%2k+WccuZ=RmS<6b-dgOV8SI%
zu#V7Tlw-)jptx~`iB+Z5g~-oOBX|908Ul1EM51O89)IR9?{_?s+oz-T)f436M(@cV
z<u*9B`A{xg*vFUgKq?fK7JO9;%ksWoV7Vtw9=*Y)a`)$B!y60?Je20lql;z)AUi6n
zs0^NAVA?10F!R^AD3+Eq3@)vgDrCQ}M~-xOg(sUCI$q%hLr*_~;U_GgkA7~K_T8As
zi|P{Q#J~ayHqRNK+qR9yUB@mb$#5PXW=iTKK4yKfwH^p4HBvXr$!T5-j*tuLQ8U+p
z6yBM%ZEsuji*kuqrah{NBg&AxyPv*P!woR<emd~o3w2D9d%$mt65zH(_!<5fTkwaH
z*b_@uWCqo=lRlIf7E9Tk8ZLL}*)dCnZsdzpuC`4mfZahQHca)sQg_vuh`6jEJNwE4
zK|Y1z3ufCAq2L>(i%5-z05ahS;S-(~1yY}Rm*WTiXfPD3$JQC-lE*RkBppV7T-<3B
zHrS@r1pY!CkeLN-CtNBW6t6=d;xshzlckU$(sICZXJlZ-r-@Fh(9n(IWKKk=zc(`_
z3TS5a-lqt|E3tKhHXZv5Mt%`*uLW%LKliv^BX>IT)ui5It~#*qeNJm|KY7Ul!NG!=
zob=mjcUd7Xw!$*R+}!E?c;<%N1};P-Cl4_8i@6@Z%a#skIU76R_EFMkRyzM{UOr4%
z>kFt%>CVWVfK*yusWdCk*elmaMHQt}B#1<y5<a>mc>b4`t*u?yM^Y=Ys=<52;n}kv
z{Q|AKJuK}s%xBN1-DS&Io^p2~nC2P=GsOe2N#PRp%|)$;O9y}zO|kqEgl_g<JH%Rt
zld$WEwU121!D(xh8#u8pP+hCg%PT--f-og8vQ~*Z?RDN9F<GfVTzXC+1BUqbWT{St
zENygWXQvi}_&^S63piUwuT**L=Vt^3S`O7gw8JOcB#abJ#IMVct))SHXCG6J){_)y
z*Lw~MG<<;%uzMq~+A`BBHRg=M=~H@c9-kDDeVIwNKUbM<eM(fQnSl?s)yT3x78#?k
ztnN+_Zz%ou#d_wb03=v(GgeQF8=2GUtQp~O@2Zp|mAqPn8rQVd@n`}Yo^C1WGloYM
zLKMaX4F{?$&4Xjow8_3|53FRdX?(3-zMs#I7N;8!f3u$qKZDlKa<G6YTl(?i-oTao
zhWL2Ff9su+^~BAMizG+#J=%$~va&>)dxno*>6Ux~H4nf;5WkifoPzxp7HzVR=$)yG
zS5<m}FM%_NR|s?pkHHQ%x7ODM8a5_NcQ9!?3a~KH=wN!$TLtA38XN#=P)1o7<T!<d
zl0ckCeG$Px#@vQ?badz-<c#p`rzL*nbG5C3f)ea2)$^D-7nS&&R#Xa~%Vv4)qk0y!
zBA9Qbi9`=A1$HU_t*M3Dy9&^pbq{(TJ~soMazF!IzjB4RyR5Uq^Tm1feW;|Au_Eh(
ze-{<aRzOcl#rpwrC~Wlz#A1*S`P<+AweJHQ9F~8ddw8(?$H4&>&hM_DKS4cDVo%9W
z@MJHA49Z@eCtW(+Dc#QEpzx+<Zr;O9AT>&j%e6w!R427Bd)>Hkqgr1(OM;&`2EO+h
zxKdEAqvM_I2a@NolT|*W>e{HnsLk`9US4n}JCM-w^4kAV1+--rnZPGg&!-kzPrOg|
zIZonD-3K#7cV>z;bffRfTg$Up%l`u45gi>J3(MV`Og@Lde}RlsrAeFSLNyRIu5V&0
zL!;tAAqcP!<IX-2kB>8gW#LRP&NvoKjbp>+VO?0&d)IxFyRAxNIUfZ>XMs}K6AQ}+
zF}}sc#oSzo|IRGePEKXxMNdcURc^fMm-!@r#1YKptIzZ)b@<K~@~tQ{Xh`8Ttz$>X
z01X9@3VZh!7m<z%<0cH^j-iHGlMRMaRMH9-pINmkFM}~5#BX~Z4YAH_3PIBv&>N=X
zt8zQu_VeL7x3}qD>4tgBpZy*Oa<7_=R#+RmK<GX?cDzbPww0E4sYa~w%IK2P6_%jG
zy^G<429e*t-rQ{^c$B<Q(-mAm05&E^Hn?tQ#%m)}IEnTqzzH)kVML|dm#Agj=IW%w
zHbJXa&;D^rZ|^_i4LHhp>%q|6K1Kzki&sl4B{Y_7mo*)=*Y-Z@)X?6fwXm=}5P~Kb
z3=*~*+pGgYl0e@tp*x<}%1C4`J}Rov{@HTd)x3`%W3J@%rMg~zV+t*Q-5d5Qv4dsP
z$6DS=U9Cn^LgF}F*l}~9`sQDe@16uwsov{ts`)!08n~prTDNJ{2(+7&0WWOtXAq;S
z-)gAx0O}3}p2?j9G*tXO$CdMhHz^B+<7~^D>T6-&?-d@^K~cYlg^xa9y<;YOIA*_^
z%ifb?U7by_K9gtKA)#LOPKQGxS0?(BfsIw7o4eGjXJ=et5q8!d@_@KMRkoP1W$WAR
ztqV5sGZKkYD{y|Z&K^orkepmIpOi#vWJBDZt6VK`f3;GyvVx0>y{FTB4E)sf8pcaW
z9|P^`larI~@O~5!=ZbAR9!u##WF7C!isnKpip=Yw{cHYi`vH8N<AvHrWxk}^u?y*>
ziy}^AwYukRZ1#2!TS88}E0Wa&zgRp}Gc9;EnAsq5>Br}pFm^H)e_EE-pY*Osh;V;>
z6{~-ia;at>1rnN50|5_Z@rS#)5x3u;`_n*w>%#R|9*!L;H+5gJFlMQ{J%^>E1=NSs
z>5l$08gpv}5vp#CTh~mrMQisOv1PFnc4_mM8m1a0Oe7CQNcxJMG4jZP-*QcMn>6BE
zH^v(!gJN(=-alKgB3&VgDU*#F4bxVLhkTZHLH0m11(|8n=6Mk>PhVZr*9K31Y893N
ze-#5<LukJ-h1@*53G3qQ?6>CREPUbDjP;c&)?$(k?mXU|smiOK0~Igm&7J}E_ZmPy
z+a9Yk2aO(DuXLT?R+nuuZBDvXg^`*H5kVEsb<Uh|ck>AuulbH9CPcEgm50^*kI?3u
z{`J#Qe_w<rd}^4TU;R*%X!~4Wf2_2!#a?&%iWkWJZZ5VK7N$Xb4Gr~N<*`_-BQfGK
z?!Q{VP*w$Gr~5WN$hvuMZ=%0^c?@UK*TB^K^?&(wBByaPNbwKCA0W*jAAl)V(2ui<
zvHq&?8e#b<ZSJS>See$q7?tKm)d#E~^>)V`hg<)nPgpzJG3~d#+4-T4mQA9b*aeff
z%|}j~^SS=r6LCk~JUZ;W7@Sy9-7o`9H{X12(s<VF#2378Y~XUc`X_BiSLwn{DW-!h
z?qQWCWKMfV?NI_&%Q9HRcWZG3>C@V&rBY*sdJwN~VBpw>Uju<cQTMaF?h4Vcl-Iok
z`9Izv$hnT=v7n1f%ICD!YhqCrW`A0>%yM#ss{~;p(nyCL^b-O>@q+7F&R2zI8eZqr
z3NzbZ0hL{%&Qs3Q6{NiQ7*W4>)xa+U5@VGei=dT{>f5{6Ak$$v!t1|d{zMPmzCbGF
z^go{R_Mb%ggq?-`=<EYsHlSQ8$@O6O>mExBWxFDqPKDyXIdc+CeKQR_=9fCXsHmcT
zu_>#am+?OCey^>qz1SR}gJ~o5DE6q<Cr(s3+2vGxHhg@rKDag6JPZQU6d(dWU}HnR
z;4-Rfk#Bwef%N0Vrw%t!)zFYO!wVN27c6}^EJHlYlC0U<+>E+PmztDx7QoCei21D4
z4=lqn^8mZ}ZeX(q6WS`>=KMe0<>URl<=C~4)RB>N>Nck&&$@QJjg<<(o(@qKje{Pa
zVAp{@)$H3RZ({?2-rPW!HaJC{zs&UKCpEuy_^WE#iZmIj2-`_J?eCbV$yJlq;i!=N
zMAPmSoRo6>0F$vU*p%1VUaK*kB+l|d^wN)Tnr@?id|_Yerr%igQ4K}Po+5`c6)a(5
z@2wNOzO<6fT#tn@FxkZpLf!XnNSw8E+d=tpdp#O*c|W*dr*7o;1lzPupvGc=*p{~<
zD)G63cZv4x&I(Iuq@<kf$_uW2RY|21M+2*{z4Udf-&3toicC(CEHa)XTA%e{L?(z4
zC9?IL;R#ECL8)vPQ+jPo`71(m{j*Mv$kQJYiWjNPfLTGyRpWp5i0@e33LzABW(VZw
zmO#NkP%ouK|4+d7yrAn$FBs5L(xcyfeJJ>RYgV+N%7=0cZrre&_$@dz#2+0^oph(-
z&6|5FeI%eF0Y+oC4{e(IEjNP7D&!ZsERB3q7?Z(23Y|Tq`}OM=cwqu#PaYlYcG@ko
z1P5l7u@!fzNx^MrD=B_TQ(!qZunO%Esj8YDc^JEq{`Qt@RXcHL;I4H}HE44>{`z&N
zjcA5?iGKJ|njns*1(OWJO(tMdR<4-&Br}?q`e(<>g~;QcnAQVb$Zhm9-1XJQoq9x2
zIA78?j}{7XKIX+KsLu;O{@(1p4)SeMlFe5cq*`1Wwv$#IhsoOK>roLv(gjZgQ-7Y`
z0e1Gzo%t%esWO_|@89y{yjp>Tu9O=jy55n7zIN@Wm2s*4Z2fX%IIV}?a^WAJ@Bg>E
zKg9kRIFE9ge76Abk$3Laf36<6uVWD`TFtXn>K>b!IR}*WR4#MlU#TZVV*Al)?L)21
zH7O`)90AIb>g{@9khTe}`XOx^9AOpam0<us<Gyw6&rK$em>D<VgfKhLpDpVbD1lRi
zyeOtjL*s_DxlhJJ*&8M11|`0LQ3z!PbsuA5+1ilS;A=1#%<l?FZ!<B5*i52sH);dH
ziWt=+_xZ9p_>uilQQtqw<weZ>JaQebek8)&^5H+P_|^q&H|JBbM?Nc;=-N24O1@m9
zoqiQ>8=jq+{A_Qc6*i0rJ6Y(OG`gKKQLv4h^8kk0wM!9A>+`4kf9=2f)6o4y@TY%x
z&B{Vyo&y;g@^N>VKfAexm0!)t<jIq70J{k#Y>5L@J>>}GxgT+G1og>(xb@Qf$)A(G
zD<ZZg;=6L;xU~Qko$XDnye#QyKW9jPRj^a>Z6u=<%fZ6Np~j7FJ*#1i2ZZ(hSyiRx
zO<232$@&YHy|Ll$s5mX|<|B_^y5)wmtMA?z5oyVbHW~Wd2{}54zOE*t38^`?_^;J{
zVK3m8T~9Dh3Q7K1F*$XLt19T~)tvvgX7u=a_zo#P|Iy)b-f|2LCYZ>i7HY@B>=yES
zW8<_pPp1|zO&|lZzxz9p%?ay->`&<v{g*7cvjdt!&cNCUiZ^6zmJIwhrfO>H+{T+4
zu~1G~yk=f^EFgIS2t(=XVF4;_w}?EX@-4GepeI<(!2Z!p*vvSYiYY2mT{Z#MSafoZ
z`(lgIljq-}mwPB@2k~&8R0b9N_Ba8UR?vCuWVz$;B5<-R4U6{ye@>cALo_%qtmVba
zi)z;g4+ftw+_^JV19Kdo2Nv^2Zd4Ck!20`WA9kthXSump_z|eBIXo;w^p1}1y&_*&
z8zu@%N=o8(fhR~qfA|~}7s0=KZX6b~v{BrF-d9kyKLw8KDo`5$Xp~S&YXFeF?dUmn
zW`H>)1ntj~@meZGgjs+9_f#5_oUI`+8No7f=QdO5N^h=?W9ThL#w<71e%rJC(j6HY
zS!O!IE}VyTDlNX{CtJMSVu6wHa>vt%C>hd#&tRsQ$WT6h)y@)ho3g;p`ZgwAj)^A-
z5Y<M6-y!c2J-YRb#UcRo7{`3zNxF1RJ(U+!7j_{v$RHiE$xvxpFtFm7Qwa=dK%AM(
z7@>FN&VE{^^NjxubSUc(yc^g^K%L<F*>Q*U0cF5XPM&yH3Jik!7ijgx;3EQNsua?$
zxBL)P%Y^=Fbd0&FdJ^#S8x{2Js<ijU1a;(t4%WOsS-W@Ewd1+QrWm>oMtVctdNjjh
z?60b59)ili8P8EqxEo2P#X9ht2)9><P1O6RU<90InVXXmEn~kB8}HhGLTuUlBqt{5
zJSERXH4KZ}NyYO%tt>p1s4L>0?49v2Mh7>}UJP`AS@(3q-AJ-vj?s-fnd_7DpFh5a
zn;r~-DQjvTf+m88He3S38Aog8pd>2{)?EuEqCW|Ed&5qJJQc>7c7bWm>x?5hLZBn{
z)0<kbWQBp!(mm?Ef~v}v)225GlcmO?M}0kjX$S`NQg&9Bmfc*f8>UfX`p-7vUT#*G
z{@5$YY2r<EG6&pku|@BPk-9nx=h0$XXC}J*NR0_JW+qG1$Vt?K&248O1UEC92Pifn
z`-6g5p^Z8&fqv3Z_G}-i%t<9_F;m!q{_0ym2-%`|1cpv*>N_7dGXm>tBNP!8wGMtd
zDSppENrWSLM>c>a9%#Iqe^iQ5G{C0DMW%*^wySz_jzN1vkDQFUKV!@87*By-1Pi0k
zT5eU;b(Ro9Hl;5q{r|{>)p6N}6XPK8FYMUHfhg5Sa)a(#^qPu;xqen90+SpsZ>t#r
zzJa&FAHV7)Inri75HrCqc(i$E=xSSFuG9EhL6x!Ch4ghE)->=+#<3o4Rc^RDX!Frx
zgS+E5twWsD!MnCb-kSrHt_UuV#?r#FrKP2&t+I0ysd0xU)N=$hD<ML+Bfow7)H2}K
z;4ZruP6#hGt_VD=;*4XIJR4YvPU?elpo(Vea4kZs`276N089{Uxf@@7Q!%$!m-Wx-
z_++U`;q)u}FtsenA#Q^6A!pePC-ezC$$>GmdS??at&WUTJ{i1Mao{Vc$l3cZ7Y5nG
zAsJA_L~&F`_ExFg?n_o<h`U!9z%up_ulZskCWy|ULYCcY)XikBQ<Q2YHx|)sfTI^g
z2*g|0c1T<}YL^4u#I&i~^*p$=z3o0HZuxwpR##JHJuhNC--om^zy%UGYviiQw~9Zi
zo#5bU)-xbuE|Wq#2<C=IN;wf!N2M*G`(h7>)DPe+T$rw|ZaOOCV*NUeH1|KS@P-l3
z>aVGCbKtsp4d=9yPhE1nUVr^M<Cv3UFQZfGiyd>D^V8k922Yng1wP7n77Kf5UkB*3
z<F{z>?H&1LU>z=s%RJf+#f-3CKxV9LZZa}3B;z!jnwo+F1Iw&hy@3zN5Owdap|Mo9
z`?qgTfjaM|`SBr0tq9-;s%T<?NeYibFMSR?Sq1c|yg)!IKWQCwjLgi=(q94WXPjxR
z=XZcG7^dqPb31`dd+^p|TI`Of^I?ZRI+?>G<OY|w1I#oiJ7AIcDJ1f}{;a6c^@Fb=
zm(zIfP}_mpAF0sO6qyvI6d(GU%vWU?bWOYPqIFe>DLAd>YF!!}2Dz92rMx_c{%a4E
zP&2TA@YcgQmoOqeZU3*;xR6C=S(s6&m8{6&d(`03)||avs&nJ51Y|I;LQ+yv{*R!;
z--&-kT+&z|?)7vtK#*m>qmv)<{C8;hf5>eXJ05Gg&6{=WVUO8WFB#TDN%K*)ennO&
zv2O;Bjywvk^V#FV7jm^dRA;>hii1S$?)y5M@&ygkDV*V#uQK5vswP2sKT=z^-k9$^
zM2%W;6iQ9;bK2Yw8pbmCBxK0`A63r0XkXyW>=-Zo>(`inrI{$<hhAS2$EpYDqwbCZ
zjnoF|Ui>?+bStMRBI_#f^q>_7#+lDh9_kAW;_i>-Qq)I*gI;1>=^hMvO!AATQx&4F
z;~OijfosoxB;q~>SZAZlTHSg=gP?t9V`6CW3T{KRG^Xc{dXW0+tIn4|WsPAXzDsFj
z&r|p7y+>WW0XA6hkAY+$Gq<Z8o5^ck&%S=B?BwLsE%FE;SZ=Pa`3{d>eF4As|55(5
z8bgW3YFgbyKiuR2fnePpR*s$)vCkk+4S?P^t&n{|Q{Mu;APc$3Qq!8XLhWpkf~Oh-
z?w4$V5!!u`=amKXTSv(UhQ^)01!=;MP9Q+hNG)T&$&|e%Dqz+0q9ys`N3Z)0t(G}b
zDxw8mtxpl4;n+MgX2)a8%<npr%nLJk_4cjzSN|g$Fi1tH%hK{T<S!JG7U{R=>%x9u
zq<!7B+dSe`+yBk+npyw#oLIs^!muRcP1SHTL`{YV$!f8q#io-VD<O-4PsKS2-^*81
z{el_Ng(QvE6TN2g1&!_$gI38F--k{y0QE{sJB>!kT+xHB^x?{nayyh|j%IG6b8MwF
zcH1FeGY(u8Ko@;qYo?gTu+(md!|Z@sNhEj8JI-)f8KLW+V&mQmfo7@?_}#<aqm*RM
z{m`T7mBNRIo1>+m`^*57m6-DJm=u6HJDL+P<xG&8L*4GL&gm!+>h}e~MOtekIlB0;
zkR4JcckNvPD}lofq`vBZwSbdAM~k3+v%s%#NP_LBd|tadLVe5oeF9u<p~Nj3^d;@|
z{iMBo6&ju4vYi*H=)vb_Z1J)5q>(~INQ)g&tS{L*Tv!@60}!%j+;6g*eM3Q8YfjKn
zkm<1jCQ$1(2Sj}O<Hr|~bFcrz|KB72D_EWOhW1>#y?G{6Axp_CW=nTX^3NXQERd)x
zP7mA57#dylJ)eu>-Rs(sNdHRqoWkPMSw~EiCM2cpw^|Dxsj~fvFEnaDoGgr>qKRRk
z{K3low`UeXdM?@x$hTr~mc$);oFA(Os7V0Gr&SpCr0jy0<9FyX7RwQyC&U&d$f$lv
zIMy$~Cq8<J^l8L`gOh{L<gvB8eo62ik}F<bau3MH&nhkC|E>x;NELOS5Mdrpha0AY
z!u^=Z@({tJ#Dl)Fcu*7UNdH1T;6@@Hy6^ZqU(iX{de|9unno@|G<apVw-?7?bAt~7
zc>ZCzOw!LjfZ;RjsBUSCal*p^qO(}*)Zx)ic6E9UrZAM&5z|!`L1*Bbnm4!$`Xw0L
zqNSy+&gp&s;6E2KcD)1weqXaTjtqOTyq@dhrP-dLr!XRE&xmkPZ42{FFGs-Y*HJY|
zV$5F7rdqUi``hED!vwHMPfZOgEJ23?G$k{<Ze=M<CaFOvrG_((9ujJ=M(r%pQZB5=
z7Ba16XK&%tE??$mt&>|mfzf$nWGp%O*pat3-Gh5{<jA`b@p3_wi1@LwF|Z+%2gZ{%
zdnO{^Sfpm6<fBt=K9sx7!qV=te!b@AjAMFKFn+_jHF)ygU7Agd0JI8hVRXcelp@_i
z_l0cf`qCCbf5#<IaE!cq>*mcG`@SW9%Vwiu{mEPZP+bfAQ_NDJ;8-ISc4m^@gg2A}
z)PjhJ2vE%=&4)iv_b|(5eEc}_MK3R$V(Esx%`Ph^9JCZrv#$D(QQkks3u$64S6s{`
z)OmOrZCqR35@N6AhN++Y5mUsb5_gqJY<|3-XIjce?C9u-tB<RjsPBS5IubMgBz*6x
zJmqkCSw}NLGl%d*0t`7FZ{yEE9e?qYwLLCS^iye4oP@O*T8((t`2!Gu|CPwtbGroQ
zb9GefuN-<a%&XHGE;YrVtoCycPA7ch6Paw+%KTg{&~}DY@EccTL5+5BWT#&g5XIF5
zmRW>}1?0!J;d8$!Sczdr;=^)z*AWJb<|!RX9}aSSOttyx>bsQ->lVa@gpo%c8<GXd
zvCRhcbUxV8kvf<{+7O;roXM!KIEegYBi#{!5O%E3&L$<WcIfLTA<JDE!+^l|FgEj)
zNK{c&yviu@Rp~=2t+(bdz4Ue>TNKS;cIN1udWc09HgMFOHL?d#6~Hsj1aCtJ2tEUM
z1ps~aXJ4Z<qdx;6VOW;31RSbNF&BQcD;V7Aqcm*X4%W256Fl{i-`W6ndjSRjaWZ+X
zGAIN?A`+BTRN}#t;#Ps-S}vS;v;~(2pJW*5(pl|KRcH6~^h^jg)VEzq?3mX03JSQv
zB?Xh6jE*$)e8(<}hjhZ$u+WX}YYdFpjKo7MPvr@4*4Cti|3`bHeRG{fj@^{>%WGLa
zkYZ`~42Wx1O}>4hp+ump_}#d7PvZH{p8*bSP#H<xG_~!&si@j_=Y<iV1$;u>vE^*(
zppl=Y7%M3yg?g|6G{H0Y;Qjdio+hO$cUA9)?6lH=$~a+3u0}WFk(JeAGN)`h?j{Z6
z45k2hGF!7zZ><tr?%uj(fB6lVNy#fU0)S7dx0@5A{&Nb#1MB}>Iy9GYxE0V2wjv-v
z?KA&#SKy4etRuHGNo)fqPzmrNL+<|1L+JlYXd9L|r^7<egkL;Q$aT9Ao|6^E!EM_u
zlU9MZB#Mdp$(9K5ygKc?#lX^#KiE^<=6!nd<y~=jpw`CnEV7N&Qh@p~zE^2{rk6lI
z$`-5bJiEKnD2S=lmi6bFi|Okxu#8icq~*n{Wuu}ReZm<ey}{B6wfmfB4a46a$9jUE
zrKgwQ`mB@qOkV)+Gwfuw5cE%d%)t{d;c8EU*B5A9g(}~DOi4LhtDQd3FA}tRslB~9
z-)b$?kvYEDE9DQmh^)dxXF=dcK!E4sw>nBiDz6QFGr+jT#R(9rP@t)W4rUmc9!zF}
zK@%|3`NwRzkA#G5HyA22f{g?%qjo`QTyMEbgZ3|R#9L14_1Njc7#4~Nfmv3->gwWG
zn^wVHS9G=M159sXUUpXYIH+4?lG$ClGMq6KluVEc{P_|oweo(o-y+MhTSlgSIFuAK
zEbG1U1r^a4ezv{6E$DoB&(ZO2F%?S2g1XUbWD|J#oO>`Z`GUu%kblq7RxU)rr|q5L
zc}(c~7Jz8Oz)c0q`qkAT_DceC6hrSoAi2ab%{dO9?hax-)}bdR0boG}txmq+EboRS
z-_NMa)8onfeisqCw6-DUH+wa@7$mu2OVKLwIe1e*uY>BcN0pL)>|L%)SMS+}Xqn!}
zO_56XT>*(7FLQeQnsQn`c*XQ$_0__b(^Ccmj7;F}H=skX5r9Ic-8Khgu}_cw6NVIO
z&b?v~dHFWD?xhx5I<?U3uddErQp1RWQP}Q6rh8bJ3b)TEA;L*p!=S?K#nFuR&o~h6
zcAMM#EW7&kk*7^Cz<_N(IabOEFP<W>zd*=V%+&<+sDf^#&$ZvDx{Wu?@}GY4tf~aJ
z3w^rxPb=+Mu3aIk!duIsO_}b|Q7=IIQqeH?22sC)O*Z;C#!kmk>Aw5V&B0yYwilY^
zPL4YSY2v|&tER5r<oW9i5Wg{ydje8___oyR%&%bd>mW|?@<ET}n{(SiU5AX%!ZHAZ
z{(rlExE>_T)L|{5XU%x#`7)!zZ?sTtL`&GK41QUvdUs5oRbHy82Wwiy^0Mtdeu=+g
z8<bd%;P8-23;l+lr?`7p!QTvfbB$XIWVd0vn^-#lm~T|rIPU0H+FXG??p*b$>C0oy
z;|wMqTS~x$1a3|jE%f%!i7(9>8QohkH^?*YnE6;Vkz()X!DR`$zfq!<Z~iX*p<C=+
zE=Ln{wb7SVPR`hAZVT;jIY4^2IV6;hRh@{A>Kbd1d<ioJ3Gk&l5u<r<t&>UNQy-^l
z$67sogPQ9Beqa=dlL}u8oO-pVG+x%zidtN3(~fWe^$}mjy2^{p@~mv1%@;368QSdb
z?<L3kW0x9XI~scR7OU@#oVVv%O>=b@a@-0$8+<_1lCsaMk@5b9r_07Y@!OcHRm>{|
zmOd%e;KIRXaULwd4O}w~KEJ;*uDk({N=k)^=iZNW*AB}C8CO?VTf00rw=Mfhe?!H<
z;9&lX816}&aiu9{utTFDCa$F(*7J6^l+~e-*Pueb`>B>(*ritTewWYpbscMo0HK}A
zoBtX%Jx!iklnVpzw3N3e0p5SNY{01TvXRS#`|o9je_2>~JbYhZj#7B3TuzQJPPvW9
zJTCGzGb<VBBQKZ(FE=KS+o*i9#F7C05LRkj6IML7nw*?`v^5*DJ+Nnn&oC~t=onmJ
z6Ioe$B|G<{qhi>+-+w_Pt@vxm-p-&|z&`%U`}bEdZ9!VDzV^v?nAB6aT|9Q~@3c9t
zcjtHhJ>EQBGM8`BRgz;RN4@8@zg?)q9Q_Z?;J)+iRKC+zoAIev94K~94%9gGNYe$x
zEBzS6(&&=o6aHm^op#!^UPb-rnrwCIk82Bir9H&&RW-VQD&25r_eb6J>v7jIYN@$R
zyC)}Qg&tTqhepwCYWB%IHd=&!8`%TzyWH1+OsCVf3q8G8`#--}Y|@yd_k*goD)X)Q
zd+zuzlP_k+Y7NfD+EX_-1}Q4OVUlklfNy9}`b`haODTNjDOF0!DjnU8dT7Uc*PFd-
zW;O21m*FM-+gDAGzu(o~!PFRu)$HFct=pO~hi?A0$1J%`hxWEViqR=HX>_Q|z^}le
zf^Q?O7F!jj^jfeNv$HI5;?Abknn6K5E;hN<t(I+pQ^$1BuD<N;&`bRoxCS|8YovRs
zUT?e}i84&M2>+P!?~O2{wZ|MiH(IJkPHE@@jwY(KaH$U22UIF@UK}e+8LJ0AmWl+L
zFq5{-(zGGrYl8e6l9(0`GcM_disZ^KEw8g%N!Fl>+sEdVD%EwOEW5C<p~cVFu(OJd
zsEv>S$po3emE~?~`=@=@%P!odnWgZH@MO&qQXA_Ke+3WuM4{~`H)3F@-l+_k?mAlF
z0^jgkBc1v^OYJab5OXtcwn|2ZQ=HH4fH|*XTI9={Gy43#n(LbjsrWR`opz7ZwqS<_
zy73+5j)&zMDg2x+TN8T`Q`41+Xlt>WxkkJ|F>2qO6BAjz$`fVL^n6~ZTp+=sJqB3J
z1{u;(z_Ql#&EIRQMFLkmcKu&<7iBZ*OA)!IxS=tNR6XXLMF_v;2yT3^O&jj)XI8sK
z&vF;$i4jwa=TL`(-u?WP!$V&fRZoT=z1nzB;Pu><(kZ6kS+umpWjUU$#w29FhQij5
zMuYDB6dW!eSzuhEFT7;-!uDCq-{mMH7e^U!uJ4-`Y7v!A{ny7bB*Ty1v^e0+@cqU`
z55^0T1w}CyW{p1aWMER=7061hH$?9naYHKEv`SZ2Vz#}INDenP9UT^@g&di%k<mV-
z^B<FwKeS-?*oX=WuaOJZdBzvg7mv50D9F+N5&r0)J&+u8^^oR~Ub(WYHRTBd!(3hz
z0o4pC(^q)?{+6AQgSQI?=@C17&HhB=UaEvyuKu+kX{QBxp&nFeP~b@~uUhj_?<-4x
z|0AJ;?QN@q6>wy8I>n-A(fvJoQbvttZcc>erbZU&7D#oew&OUFd}C3Lob(`$X>oaF
zllNw?cfz;4{l4#`#VQyZBq}97`h7yk>d)UMjm~^DItZWi9$HF+cSKFjjA^TJ#9?i%
z0q6BDMHMyIROOPg@0N2@{bKj~O+RZlSVh=}*>@xl+!u95e`93K=3Aw@iB=n}!8|q<
z(9@e{PeIZ0Th@QBN!IZ7eY01~q4v+h4ovr1b%<t=0J?j{GXHz<+C*Y!=ig3gRB&FN
zYlpEeX;u>9>h*|$u9?QL<<zk(cvZDq8b3r$ZlO44i;pd^85RhNy)_qAt#-#EYvjVn
z-F%NXH&$JK<?chNuX5M9%rz+x1C%YAxpkZ-+f*hu{5U3<<{K<i{ZB3%g`S=FcCa+X
zj`gNq!|swTX}4l8@EBI`Ur`ro-&F7QyP-y#z!9{za%0>jmET-VOiZo40RBK&pr89t
z-LtprKytdd#UZ{;!A~mNbzps%F>R+g<l5>Len~~+^t|L6nb66xAlh5T=Rr;1zvt8x
zpBfg~M;FhFL3pmHlL$sqerE}acTH6+gRk%I?x#wyTYJKFQXS^*9h42^hZL_iDE$a}
zk$TNYKVkFSY_m_^d|D)?n0O&6aV3CDP)=Kn0UZO?yBEJzyWgDRcD+0VKj1FBuay?7
zxF*bXP_DploxnGKdTNv!j<&V??Crij)t+B)`_?_f?!`r`9DeZ=XnorgZ<5dh0f{JZ
z%UrzAKOe+Fo4S)4xD-gkRCD%Dk|z6GH9hyf*Vb1rg$`knrfgbN=pI)_9u5^Wg6~>l
z3S%f|Z$fg<_3g#BMG19o+6!-GAHXmHuBORp(|cDQ2~rv=@#$NS188=_BDz&SjJR&`
z-2alYNtO?Jp4O7))r<ToJZ7JEx)Pm7<lw7yo#T_)M3o@J>8-WY?|}PoJkS+$`%3NU
zR*JJGD%ZDBEk62YpU3zS1RSKB*NeQ__q2lr?))of;1Uwms~y+e;PGx?v;WsT)=r^E
z^Kk#j`gol{ce8YPtlN3K$}u`=HQ!W#FU;w>diHl;#R`#QDSsSi-FbtR<VXRIy1XcV
z$F^x|>eR7I3vJL?5ohEXht{_do-|Eife(N^e_DV}eOS;ORX7lnvM=PYHsHmz2j$at
zVEao4Wy`fD)s(=?eXa8Xtp4Nc*a=7HNjKi7f*cKtF1DZ9^bLfqg!#p~4SV4k3Qu&4
z%lxpXv1V#B*X6=d!BvpUy4SB?FHIvsQA%4<7)~wy@xT4(MGp3l%w;Gl<8s%zhSlfH
z?!+02t#l#(%1+914ZiCn-hX+a+6FmcAEUtjum9$(&a;JFlC+_NMT#h%fT)R${Yio~
z_-`*-*Ei%v9ra#Oznt{ymrZVgfBmLy0!>{Q{sFvk09o|FMIH6_l(FjR>?x?@hby;W
z87<MK&ozY_s^3*74L8|;AAXyO>wY*E(ok1t+3q}AH*h|qY`&meN^K-*C%Yxmb#h;E
z9JbojXlAM_PO<nc<=<@mV{8L@z{VF4^Z;jgdDJ1?k$2F)LYOOvSBkqj1i3*)E=^K9
z>F=7<y8MH?=F(z|)^wSu`HAS)@R1I`4RP}t8N1ar5@8Sp8Qv7j6AE_?aKm{1+N6ey
zPlcnww>apQH+KGL&?Sc8V|l+ZjP9><kJ@G+0ok>I<g?ZOuJFM`huHi?aB!%aKP9(U
zn&su7u~0(2Vwt*;3;i5I-#ZaI=%3|zG%;Gv0Lh@&*Z+{Purl@rwT#P>_xUl=5*Hht
z(z=|p*yiA<#+_At#X_CmRDOcRM=n_`d0xPvJ}EDG{FSV^R^>xM9hA#ouQ)gyGjbk8
zkNR4__tOs%F8Cf}suhz)|1SAMsnhvZE0!B!t?4{0>JL*D?dd^82`Myd#J@lNX}aG$
z)2w3OW^0RSC9ZirwjuS-NfjGRi$;MzyP>K|klFhopZp0A%b>`?*py5fk{0UvQvI62
z`Dm|qDNi*?Jq~PJ8hw+YRB?jR#?AuL4P+Pfh_TJoJB{;uz11oEN<ZEzqRUjcG{RKX
z)Dj=W?B()O%Ea$N0+v0gFX-D3nW>6wD?U!t<e0o_K!2<H*{VnJ*zJK~HF$<vbhIgP
z@oONtwhUTqRvII^YFzEx2!7=s#pF$AdC~?fkJnvmOqHPSv`MNQ{d6p-^O$x=jUkaD
zp`*KuX~&$xt|5=GX79DD-57Y18~H!CBovyfN6^(|m`+ugAX6>Z-g53V`*pl8&0Qn@
zRo~XQ*7k{<`Jp~LxR0Rmx&kS#pB`#~4}1P*Q>t&rb^SCuX_8em#Pn%*l)ik*s@?qE
z`$0NOgTDqW<(2nE)Ar4D`-H0dO3%*PR&k{SOd0)t&Lt!FcGrvkavD&9>5ECYv>WGM
zE`7M1tJHISr*ygqdZIoV=cvZ@kc!}bK=X)8w1K{k-s)N_@3mWKCpRoIgcN~-l&#_J
zKH3uIT#tFlF31*Nnf!6}j317EP+KF~KDE-7m&Z0frau@5cP?y_+@ZEZf1jJS{=9~N
zDlS@lZDZ|Hy-fA!!c6r2Ys~{@N~t}jh@2ou8Z@}LS^X$Ma&LdH0gy(lVq)Um>Mbc^
z71-Rph#7}46FcRm)YYr}@Ea0&S_EWqQ#ir6m<x}US5I20`@+K+$NZT2a;AZ;?K5X*
zXIIUX>L>5tj9r3fEw8Tf?9E9en>vc_<bPcLg$+Gf0#fX%owCv!y6JZ)iKpASWp+^E
zgp6`taddzu-&sL*l8C;RhKYAciuH<v%mE*dBLq?-41`-S4#;@VkE7ofIxy?;M)JZF
zYb%Rn=IY!mOZ{b=2tkim_9MXzODkHleWO>se(tAFA##OBkd2RxZYp6l=u}bJwYUEP
z>H`?Rc<hF=q`19mmA5Xfa)Z#Iq{JagiJC9CRHv%;&u1>wek(Sk0ek|iw}o+}LHc;z
z*w|?P8CUZYZ8dwhk+1Nod;#;=vlKO~`O$9C0Gr;b2ZBm@B}4`7#y-A6BJf#T%d)jv
zRql733YfowTE<M>eV4*g9{%S*nEL0<9kmIyn0Oy=Z!u}(rOwYmHAqValh3vBmEmAQ
z2UjqbSxF7g)DM+tMOoeEhNP|K)m}mR-MdZ4@AmtAi#D^vjG67I>5#CPII`)1&Cbu6
znjKvyf`$@Ky|;^vDxeUF^xFt)Z@So&>1gh3Fb1%my85(u>sYDS4KdfM&;#ch*-?#C
z(pHpds@2qRyJ8_TO;Jq8*pQV(0C&jxLsJ2;!x&%8(YW$qx>k9r!s4SJy<M`UsR?I6
zcP+oq<zjC}`D%eV5&grO-MFUTMXIK}=(i00zTk?7#jCG}5jhm1xKYXE(cTrXAbM{F
zYiTn}Kf-&~-0Yt%HbE+b(MBg*!Z24~G`$a3*syAIV?j#oV`<u8T7m$zI<4h+Dic^q
zxayi+VRSKN9u+|rx=EABO)l7Lc3ei>=pYs>u-*nIh%JkDgKE}3a}86I^ESmotE+_r
zAHPk8W3s~`XW7C~y6~WVd~*t+((1bj1<I4(@TKXAH&N*PIo+}w|5Wl=)+cQaVHzAi
z6svN?>Olf%ieV<f?>Cv{d^X<Aue5FE@66|ggQ=Fr9y_fK6}!xFjaj-SpeHq4>%tZj
z<33R;=*Hc<22Gk5i`}7a<6pmir6b-Dz4IVnJsq#f5LD#OUw8Y4(;ZmwiRwNzc$7+X
zj=idMplcp^EPY_w=O|*OuD$U-yGec|Uny#wGtNY6el%&Lj$vkIhAc}Uc2Z8QMVT<A
zLK!dVj4nzyXSPXPkJFwuu()0=srGbidZ)lA)_xomHZ-4F_iz?KleVJwGtfDY62AqW
zl?XBCWu`pxNg@~&vPWD@-MxIR@r8eSr=nwuCh07qdMN%@EkJe4csqcC5&fW>*saaW
zv6H^luks^$;0gM&5x=F(I*Z9z;M8Z-+=HalhH|%hs8zz^mkO&!`^}9$bQivjoM-<Q
z!bcKbx1SQ9mf<fdXE0XP#!Y$z)}%7AvHhrsT+KVoNJ9_x2W%WhE4WGHQ$5td_77Hg
zO)53|c3rh->H_TuAJp{D7+)B^tkN3S?z(2h7jCT&kGg0}>IXghRo#+rrW>0Yr=3R1
z%wkUqSYtEl{X9KueS=;eS)Qt6hYv>rJJZT7k6yaXH)PiZB!5bR))+G#97_dx{hC*k
zh7p5f%^X52$qGFh!rI$Z2x$&J89J8(YqoB5S9$Q(+dZ$U6DjpC&ts&2*7(OcOJ(y4
z#5krwvxC&GAp4C)v~)`}Bf}z?g+$+R@Nzut)TkI=ZQp<gX_m3#;P+5QWe!b<l#h@6
zK7OKwYAOJ=Vk;}?-fp-@_UOIArc20H{@cvy3gZdE*UVL6;>Q;J**DcaQRGfu9tia4
z9J^g8DMC{rKW?Is@@V|u2sv3?mW2dtY;?5PAzQhCmG&`npl4Rd-*7aWwy&WgoZb%z
zGck71@;Y~n56}_1)wB2V=yVJX(n<XO?H;Olvum1_4l<V^^xF>dajxebwVA1T^^}dX
zTD+$&>sNeGt=D%7mzJ`kECmLir1Bf+t*UID4zjDKtH(uCQx`iGi<yqqX?6E~_Aj8R
z;Pt3z<bTB07nPJcU2$gK7O*8nNPO~ne;DPuzAEiUo>%L81?4e>Lt*3rYJ+FZ2`4VP
z4K_aGg7MxeQw~6FiKRIpH`|j+)7&zvI!{mQ$o&Q0uIr6qYCb5xS1YH`zt4R-{`@R(
zA=-|L8&bNEuSEPti?sXvz`|kyypJ5PRUKyO7e}7;$6xO`#q~1VI?=29X!(EBfG-~>
zZe}yyn~MXPe9#;;QCKM5+yv(}(w^lx3C(VifNhx6(c{5*yZH;tbo&PX78VFs1?-#%
z+3eBEyiG0nuK~@qf%*1cduL8TmE4og8f<<GC6P+uoiCJW1oi(;P_C)sya$$C13M`J
zcBd?f;7`u6O{A!p_K#lKG3-u6bz!|bX<n~C{#HHVwG6&0;j>X>m^CjW13h2-)Kp$=
zfR`Gv%439^bx9}-b(%=!HqbL`Fzsf@U*z$7%oU;d#v0zFi}i+ouC}n?=g4!+E{f6k
z_!|#-_Ifp9XRZ$E+K6ImI+6}Ld#<jZB$C~qTFEO~o199+CNkR&(@6J*|CwOd*f%$z
z#bVNn7OM#vF)fB^ITwa=q#p2?&33{|Clh}Z6lbP?`N9J)nQ)zNX?3;ajAxPv-G9g1
zp*%RN-uH;D1uOL64X=2yHC<tvRQ;{j!5I0ol|2O^3P*oR8W+vhoDFZL%9|*ZdWzPg
z`(N@?{#v%RV^{18yXn-{Uzoa8{%kcP7sB7CQhyk&oh@O}95xPw3=OORPBN>J{{h!C
z|Lwge1R`^aCXbiuot-)t$OQ^+;Gw{Z2A{xO&CW>w_w^N<=Y2x@wzE|y!3%Pm&Z^*<
zXI{5+gMq2halODo9D=*TUHZGoC39LJ?a54q0Rm`H13LkO|M~iH9I#1#amGu<X%Cs+
z{Cc+8+mNSCa@a?8U-o$tJyva|)swPD3g|n92+WDuVLL5xxam`YOjK3&`YFdC*Z)pj
z8a~sk1Kc>jf1pvG%7P=$clYpVOdE@B<gvhoTK{5XUJk#t45xH^KcvK}L^BBO!+4jW
zH5eUYClG?JP<ZrL2Y6}%-8W`hda=n1+BddpU{Go8%R^IXx94nzRt0x>VwXYkeJuD2
zkOBu63^HD3PVosdojy8Y=ios4qAzm@V|e8F-49No1r`5SdsiBkWV(il$})Fz#?o;s
zC8fl2#4_9}ml1abDHR!~6g9I_Trf?W$(hn<QP9fB6;M+v$psD5DM>Lkmn4^T%B3+x
z&E2$|7v{{l&YT}}mg}5zoged$AHWCS_rCA5-1l?e8f-uFmVNng4IyD+w|N7@l>B;K
zL(I{(WpNK?{(W=o81cyM>HdgSJgyG(9nqta^4m-Cv^wpr)|N-XfxWWU5e@ArkB{f+
zjelAunGu{r61QepE>BjUGR~@qnm>1d@o_CWW@%9a^}yac#^boL-w$qngi$mPeH4?<
z?mD9&8swZ51X{5#WkMkGZsJ}5Ey4ICXbgLVq;Y8Es~K{Z31v<kbw}86$NC=l3KUJf
zdl*~fFBI%bznayaFqE*<7Iw9pv&$m5SItN<dg7V1tC7}_skYG8`-Zvz(J}{y#N0BB
zZf8)QTHQNeY-F(4;X%P+1~XKX=a|J{@ONruj0DC7fF7p|(Y4rzyKn9h;>6wWeW_Qq
zn|*t`J4VwDZA`-p&os|_<zjjg>pZRre#}4r*=OXBA7aW10`h2-9YgRFdg!Ni_M(CN
zuYb<5;B84uP2DBKI#zgzrZ1WXEtigM8CSMXYGR6tF)@JR^HIi2Ud749Wf!(0H?AT7
zJ$+7*@2h%k-NnYn;{#hWZIIGm)l`(Ejq8Z!ios%;cW$!FJ-GNnF}Rbsi=xa3Xk?Z*
zJ83W)-x?pA*S&GX0fjh8jE^X<%`S_l>23z?JWR32<8GemqYS%y#(kW#tV_oh(ri#<
zI(HRs$op#4rgq;npTYOKic2TZ422n)u@mK>S3;S)>sN8Y?(C-Uvz^#5;a&luBD9ag
z7dhUXsYLv`e#!NbVn>SPvB-^)0KNk9At|7Nh{oJro>Arabt%<ZFeO}VGlX;<x7e%c
z!HK6SgYUPmn5?aG&Zpw4L-h>Q(xd>UGNBAi6>M938p>l^q`<+&uB^059hn{k$Uqf)
z-%Ty+Ydf*bgLJ8uP|^6_5}&~NT(Jj5)%F5KsD3!>vKIWwP8u-G%)X}x%oP3>p%HsE
zKp@Zc#k6W5p(Nov{{xbE-K&Y=*&y~~;N*aT{Dr6tTqL%MrL`c5c1jy(fRSL13zw;r
z(Mph(HkA(^T(W$Ws0?a8%Eikj@`cz+2L>Ez!HqW}m{$fa5#D`19ETELUY}&Ti42DG
z2&@Ba02MhM8Xp(wRou{&T1Nw}L?x}f(Z>2$_6j2RO4!>z$^Xd7J9yX{B&2O0o<Hui
z{}k3&1qbDUSn>zba89gT8te)V<B+bf;gKQYb0N5(YTT#Uq=+4qJnssjU$s-iNJ_%V
zm-&$wJ4T|y@NeI0Vao0Boe>HXuWF+Oup7k(8Sfs&t-3EIzvzv6N1tnmo_rY-vno#`
zD+Y6tu@>Ao<u^mac$X}5WhKp!n@;?TNTEq7hQZ9tw8spWf62Q1g~bCL%d^ENx@jpY
zoQ0mN28saDGgD|+*Ob5FGn+tVnJrleGPUYAfdfVV#}JxRUzEA~XL}_O+R>sH#e>y{
zTv2C!aI7ch<!wW!BM6LGy3AV_Jv~_COXHE9%Zux&h6G?x=?>e-iPJF|&h6J-q+6bj
z5me0K5Cl-Zr#As0PJZH5`t|GD{zhhIO;-1YE1d6K`Jz95j^EHl9R}H2cfl)dB+@oF
z(AzsC35^^^UeV|w?%sk4R2%O-?lw|X4*F05P#;tch$Kski&>}$bJ%|P+dgce7t6M<
z6l7`}7EYujZ!~f*x@F^hL)*dvzXozjuv0|T`rKz~GZq%*WQtlzOQ5Sj`Zb{abE<b4
zg>URqez=6L-A^9>X+k6%xol1%ku-OYD-}Kxy)5JHJ4dI#Dix_CZdv%p0x2RM_4NHc
z%57y_?{JS?+1aVt^ySz7xJyXurZY-{QnW_f;r`7aD}48|B&$o4!f5P`%v>i$^Jd4@
zgEk>g8Ytl4a>C`b9$Ax<Zk^n*<T()3lRGFY#WjBz^ZF7-2~-ReE%DNVY_d){Hzh7$
zzO*2b!hw8d#6s2SXBSo8moe+1yMZX}d}zz`w|q<X2DOD}{rYgP!mY6*sRnYJqzl5?
zv)_7@Qd0vNMM#h#<#!e~UbrX96o~brAXV$dwrc{b0uV(2D&+@|!&U^G%_OSb5eTvV
z`aZR25T&^BD-YvVL?ThZ1?SP4pTSg`zF%#O<)l&MsY%n-#8kt+s5|a=UcY$W?4k2a
z4JQ7@*%097*+=>rn-e}a1teh3uM<BWxbZ`5jb@Qs4{G21I11x2$oL5i@FO(Ms!)c8
zC7uh*cdv8@G`0x~0Z?i?6NoFoY`<p{tVLK2K?;-ba!0r%zmSk*$<~Ip`)ZRGrczB}
z#-|1}d_E!N9z>sVOG5Ln0Ces+4k0V4o;v#kCf(VB9WdD9hxNzelj(GB7tpxc9JG5&
z>#hCU_HvApp0-+i07&gr-M|U)ln7*#O@t~_FlKC7cK?uj9w-k0*Sf^2`Q;5(3Eppp
zeIIh2N*clgx6#{?Q0e3~2LO+wFi6egPFUK3?V1bkWlWZ%L^ZRtADLtRp-vC@Z7bq0
zwRXAbsIsSBsoh=a?~+i2!JZ76@AGwc0Pq1@<n5Sz2IO^z5F=p5-6>lj^r@UAPt5M2
zo#ua&eJ9htTN;3wz5u$F$1mUBe^%ejp#5<(%Qp$`pOZnMQ%B=!K=xa>p|`h}|4bMf
z%EuPKk-Q=Md7uMnk|>meS}V2tCb<>+nXY;t4l1dp2u*@UUypImwg6I9zHe4})X1Rw
zwb$U-NWGOVd~$z}`!pgpzWQO(`L?buDnRQ3=+<beqv|KP!(EW#8dEYVCrd>_oFSb!
zCIUxR4e=>2F}uU~G^pMIlPKMA1*v1{B{>`qi~&V5pcQ8A9wPBmY9&EV8U4y2u)+3E
zb;^owlbVFF_Pg_2^5HxQ5df0y{%=HxKL{{X3Rp_|-xvf-`KW&fO*}(gdD5gnhkq|I
z>1ou{71GD00h{1R>%7!bj%S0Sg!jo=)a8D8PO*;bI8XlFyVW1!brn;k*Yu~zN}a?N
zDuy?LEtY6@?etji*at|o*zU<Ko26suY-B|$+9HlniQ8Cv3p{lUEK7rdw)sKGg38j7
znJ5Po+02$;o-h!qp`jtZV@i(BZ5NUlaiMS0BFAFs(8M=wO#^P-J3e-lcNHvtLx$!^
zzN`<1Xv!9-^OuPE^-omfJ)v$7mpF_KWF}#wBUcT-OW|{52dAb6EQQ3rhIT)kLE}LE
zfGMH}w#PO=MOIElc8!Wcl&O^JhVR(lg-3L;@|jt(4i3(q4Qi5^zBMlM`xR8xqlxNE
z!XVdjMeFxhZxvl}c>Waj?x^ydgTca&4(-1^9+V{LJ=gsg0{I_2rhg_6`19l9{qvvR
zAb;NZ|JjJ*Z(Ha8b<2EeJF5Y>*bvCZ=}OJvGVqdl9*p{ajTYzd|Am)uaH-W*Tr8lA
fK_F7%+Hx&~BRm(QFt%Qjo;%pO*pwdp`rBUs>BTy9

literal 0
HcmV?d00001


From 0788d4e52ace795d7e7f6915bf9d3bb973e43651 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Sat, 7 Mar 2026 00:04:25 +0530
Subject: [PATCH 09/15] fix: scope permissions, bump plugin version to 2.0.0,
 remove dead save-artifact.js

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 .claude-plugin/plugin.json  |  2 +-
 .claude/settings.local.json |  5 ++-
 README.md                   | 11 ++++-
 scripts/save-artifact.js    | 72 ------------------------------
 tests/save-artifact.test.js | 87 -------------------------------------
 5 files changed, 14 insertions(+), 163 deletions(-)
 delete mode 100644 scripts/save-artifact.js
 delete mode 100644 tests/save-artifact.test.js

diff --git a/.claude-plugin/plugin.json b/.claude-plugin/plugin.json
index bb04ccb..180c8a0 100644
--- a/.claude-plugin/plugin.json
+++ b/.claude-plugin/plugin.json
@@ -1,6 +1,6 @@
 {
   "name": "qa-toolkit",
-  "version": "1.1.0",
+  "version": "2.0.0",
   "description": "Claude Code plugin for QA testers. Structured PR reviews, test case generation, bug reports, regression planning, and 6 more QA workflows. Auto-detects your tech stack. No code required.",
   "author": {
     "name": "QA Skills"
diff --git a/.claude/settings.local.json b/.claude/settings.local.json
index a2c4244..e945dca 100644
--- a/.claude/settings.local.json
+++ b/.claude/settings.local.json
@@ -2,8 +2,9 @@
   "permissions": {
     "allow": [
       "Bash(git:*)",
-      "Skill(gsd:map-codebase)",
-      "Bash(node:*)"
+      "Bash(node scripts/detect-project.js:*)",
+      "Bash(node scripts/session-hook.js:*)",
+      "Bash(node scripts/state-manager.js:*)"
     ]
   }
 }
diff --git a/README.md b/README.md
index f74ed91..511d810 100644
--- a/README.md
+++ b/README.md
@@ -163,7 +163,16 @@ qa-artifacts/
 
 ## Permissions & Side Effects
 
-The plugin ships a `.claude/settings.local.json` that pre-approves `Bash(git:*)` for git operations used by the PR review skill. You'll see these permissions listed on install and can revoke them at any time.
+The plugin ships a `.claude/settings.local.json` that pre-approves the following permissions:
+
+| Permission | Used by |
+|------------|---------|
+| `Bash(git:*)` | PR review skill — reads diff and log |
+| `Bash(node scripts/detect-project.js:*)` | SessionStart hook — project detection |
+| `Bash(node scripts/session-hook.js:*)` | Stop hook — session archiving and activity log |
+| `Bash(node scripts/state-manager.js:*)` | All skills — reading and writing QA state |
+
+You'll see these listed when you install the plugin and can revoke them at any time.
 
 **Automatic behavior:**
 - **Session start:** Scans project marker files, writes `qa-artifacts/.qa-config.json`. No network calls.
diff --git a/scripts/save-artifact.js b/scripts/save-artifact.js
deleted file mode 100644
index d0ad467..0000000
--- a/scripts/save-artifact.js
+++ /dev/null
@@ -1,72 +0,0 @@
-#!/usr/bin/env node
-
-/**
- * QA Toolkit — Artifact Persistence Script
- * 
- * Runs on Stop to log completed QA task metadata
- * for activity tracking and metrics reporting.
- */
-
-const fs = require('fs');
-const path = require('path');
-
-const projectRoot = process.cwd();
-
-// Read plugin settings
-const pluginRoot = process.env.CLAUDE_PLUGIN_ROOT || path.join(__dirname, '..');
-let outputDir = 'qa-artifacts';
-try {
-    const settings = JSON.parse(fs.readFileSync(path.join(pluginRoot, 'settings.json'), 'utf8'));
-    if (settings.outputDir) {
-        if (path.isAbsolute(settings.outputDir) || settings.outputDir.includes('..')) {
-            console.error('Error: outputDir must be a relative path without ".." components');
-            process.exit(1);
-        }
-        outputDir = settings.outputDir;
-    }
-    if (settings.hooksEnabled === false) process.exit(0);
-} catch { /* use default */ }
-
-const configDir = path.join(projectRoot, outputDir);
-const logPath = path.join(configDir, '.qa-activity.log');
-
-// Ensure directory exists
-if (!fs.existsSync(configDir)) {
-    fs.mkdirSync(configDir, { recursive: true });
-}
-
-// Scan for recently modified artifacts (last 5 minutes)
-const recentFiles = [];
-const cutoff = Date.now() - 5 * 60 * 1000;
-
-function walkDir(dir, cutoff, baseDir) {
-    let entries;
-    try {
-        entries = fs.readdirSync(dir, { withFileTypes: true });
-    } catch { return; }
-    for (const entry of entries) {
-        if (entry.name.startsWith('.')) continue;
-        const fullPath = path.join(dir, entry.name);
-        if (entry.isDirectory()) {
-            walkDir(fullPath, cutoff, baseDir);
-        } else if (entry.isFile()) {
-            try {
-                const stat = fs.statSync(fullPath);
-                if (stat.mtimeMs > cutoff) {
-                    recentFiles.push(path.relative(baseDir, fullPath));
-                }
-            } catch { /* skip unreadable files */ }
-        }
-    }
-}
-
-walkDir(configDir, cutoff, configDir);
-
-// Log task completion with details
-const entry = {
-    timestamp: new Date().toISOString(),
-    artifacts: recentFiles.length > 0 ? recentFiles : ['(no new artifacts detected)']
-};
-
-const logLine = `[${entry.timestamp}] Task completed | Artifacts: ${entry.artifacts.join(', ')}\n`;
-fs.appendFileSync(logPath, logLine);
diff --git a/tests/save-artifact.test.js b/tests/save-artifact.test.js
deleted file mode 100644
index fcddcf1..0000000
--- a/tests/save-artifact.test.js
+++ /dev/null
@@ -1,87 +0,0 @@
-const { describe, it, before, after } = require('node:test');
-const assert = require('node:assert/strict');
-const fs = require('fs');
-const path = require('path');
-const { execFileSync } = require('child_process');
-
-const scriptPath = path.join(__dirname, '..', 'scripts', 'save-artifact.js');
-const pluginRoot = path.join(__dirname, '..');
-
-function runSaveArtifact(cwd) {
-    return execFileSync('node', [scriptPath], {
-        cwd,
-        env: { ...process.env, CLAUDE_PLUGIN_ROOT: pluginRoot },
-        encoding: 'utf8',
-    });
-}
-
-describe('save-artifact.js', () => {
-    let tmpDir;
-
-    before(() => {
-        tmpDir = fs.mkdtempSync(path.join(require('os').tmpdir(), 'qa-toolkit-save-test-'));
-    });
-
-    after(() => {
-        fs.rmSync(tmpDir, { recursive: true, force: true });
-    });
-
-    it('creates activity log file', () => {
-        const projectDir = path.join(tmpDir, 'log-project');
-        fs.mkdirSync(projectDir, { recursive: true });
-
-        runSaveArtifact(projectDir);
-
-        const logPath = path.join(projectDir, 'qa-artifacts', '.qa-activity.log');
-        assert.ok(fs.existsSync(logPath));
-    });
-
-    it('logs timestamp and artifact info', () => {
-        const projectDir = path.join(tmpDir, 'log-content');
-        fs.mkdirSync(projectDir, { recursive: true });
-
-        runSaveArtifact(projectDir);
-
-        const logPath = path.join(projectDir, 'qa-artifacts', '.qa-activity.log');
-        const content = fs.readFileSync(logPath, 'utf8');
-        assert.ok(content.includes('Task completed'));
-        assert.ok(content.includes('Artifacts:'));
-    });
-
-    it('detects recently modified artifacts', () => {
-        const projectDir = path.join(tmpDir, 'recent-artifacts');
-        const bugDir = path.join(projectDir, 'qa-artifacts', 'bug-reports');
-        fs.mkdirSync(bugDir, { recursive: true });
-        fs.writeFileSync(path.join(bugDir, 'bug-2026-02-23-test.md'), '# Bug');
-
-        runSaveArtifact(projectDir);
-
-        const logPath = path.join(projectDir, 'qa-artifacts', '.qa-activity.log');
-        const content = fs.readFileSync(logPath, 'utf8');
-        assert.ok(content.includes('bug-reports/bug-2026-02-23-test.md'));
-    });
-
-    it('reports no artifacts when none are recent', () => {
-        const projectDir = path.join(tmpDir, 'no-artifacts');
-        fs.mkdirSync(projectDir, { recursive: true });
-
-        runSaveArtifact(projectDir);
-
-        const logPath = path.join(projectDir, 'qa-artifacts', '.qa-activity.log');
-        const content = fs.readFileSync(logPath, 'utf8');
-        assert.ok(content.includes('(no new artifacts detected)'));
-    });
-
-    it('appends to existing log', () => {
-        const projectDir = path.join(tmpDir, 'append-log');
-        fs.mkdirSync(projectDir, { recursive: true });
-
-        runSaveArtifact(projectDir);
-        runSaveArtifact(projectDir);
-
-        const logPath = path.join(projectDir, 'qa-artifacts', '.qa-activity.log');
-        const content = fs.readFileSync(logPath, 'utf8');
-        const lines = content.trim().split('\n');
-        assert.equal(lines.length, 2);
-    });
-});

From a84f61e26f636c1576b585b347673b28db9523ed Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Sat, 7 Mar 2026 00:40:53 +0530
Subject: [PATCH 10/15] fix: remove legacy .qa-config.json dual-write and all
 stale references

- Remove legacyConfigPath, legacyConfig write from detect-project.js
- Drop .qa-config.json fallback from cold-start-pattern.md
- Update all docs (README, CLAUDE.md, CONTRIBUTING.md) to .qa-context.json
- Fix stale save-artifact.js reference in hooks-toggle.test.js -> session-hook.js
- Update test assertions and helper to read .qa-context.json
- All 22 tests passing

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 CLAUDE.md                                     |   2 +-
 CONTRIBUTING.md                               |   8 +-
 README.md                                     |  23 +-
 docs/index.html                               | 267 +++--
 .../2026-02-23-plugin-improvements-design.md  |  58 --
 docs/plans/2026-02-23-plugin-improvements.md  | 667 -------------
 .../2026-03-06-docs-github-pages-design.md    | 131 ---
 .../2026-03-06-docs-github-pages-plan.md      | 916 ------------------
 docs/style.css                                | 640 ++++++++++--
 scripts/detect-project.js                     |  10 -
 .../shared/references/cold-start-pattern.md   |   7 +-
 tests/detect-project.test.js                  |  10 +-
 tests/hooks-toggle.test.js                    |   8 +-
 13 files changed, 780 insertions(+), 1967 deletions(-)
 delete mode 100644 docs/plans/2026-02-23-plugin-improvements-design.md
 delete mode 100644 docs/plans/2026-02-23-plugin-improvements.md
 delete mode 100644 docs/plans/2026-03-06-docs-github-pages-design.md
 delete mode 100644 docs/plans/2026-03-06-docs-github-pages-plan.md

diff --git a/CLAUDE.md b/CLAUDE.md
index 9968fde..4ed2970 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -23,7 +23,7 @@ skills/<name>/scripts/      — Helper scripts (e.g., Playwright runner)
 ### Skills
 - `SKILL.md` must have YAML frontmatter with `name` and `description` (one sentence)
 - Skills produce one-shot structured output and save to the configured output directory
-- Each skill reads `<outputDir>/.qa-config.json` for project context
+- Each skill reads `<outputDir>/.qa-context.json` for project context
 - Reference files contain non-obvious, specific guidance — not generic info
 - Include a "Suggested Next Steps" section with conditional cross-references to related skills
 
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 3d2cd98..bdbb02a 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -8,7 +8,7 @@ QA Toolkit is a Claude Code plugin. It has three moving parts:
 - **Agents** (`agents/<name>.md`) — conversational personas for multi-turn QA work
 - **Scripts** (`scripts/`) — Node.js hooks that run on session start/end (auto-detection, activity logging)
 
-All skills follow the **state-aware pattern**: read project context from `.qa-config.json` → adapt output to the detected stack → write findings back so the next skill has more to work with.
+All skills follow the **state-aware pattern**: read project context from `.qa-context.json` → adapt output to the detected stack → write findings back so the next skill has more to work with.
 
 ## How to Contribute
 
@@ -34,15 +34,15 @@ skills/<skill-name>/
 
 - YAML frontmatter must have `name` and `description` (one sentence max)
 - Begin with the context preamble: load `skills/shared/references/context-preamble.md` to read project state
-- Adapt output based on detected `languages`, `frameworks`, `testFrameworks` from `.qa-config.json`
-- Save artifacts to `<outputDir>/<category>/` (read `outputDir` from `.qa-config.json`)
+- Adapt output based on detected `languages`, `frameworks`, `testFrameworks` from `.qa-context.json`
+- Save artifacts to `<outputDir>/<category>/` (read `outputDir` from `.qa-context.json`)
 - End with a "Suggested Next Steps" section with conditional cross-references to related skills
 
 ### State-aware pattern
 
 Every skill should follow this flow:
 
-1. **Read state** — check if `.qa-config.json` exists; if yes, load it for stack context
+1. **Read state** — check if `.qa-context.json` exists; if yes, load it for stack context
 2. **Adapt output** — use detected frameworks to shape format (e.g., Jest vs Pytest syntax, Jira vs GitHub Issues format)
 3. **Write findings** — append relevant findings to a skill-specific log in `<outputDir>/` so subsequent skills have more context
 
diff --git a/README.md b/README.md
index 511d810..21d896d 100644
--- a/README.md
+++ b/README.md
@@ -41,7 +41,7 @@ claude plugin add github:cyberwalk3r/qa-toolkit
 
 **2. Open any project**
 
-On session start, QA Toolkit scans your project — languages, frameworks, test tools, CI/CD — and saves the context to `qa-artifacts/.qa-config.json`. No prompts. No setup.
+On session start, QA Toolkit scans your project — languages, frameworks, test tools, CI/CD — and saves the context to `qa-artifacts/.qa-context.json`. No prompts. No setup.
 
 **3. Describe what you need**
 ```
@@ -123,7 +123,7 @@ Everything saves to `qa-artifacts/` in your project root (configurable via `sett
 
 ```
 qa-artifacts/
-├── .qa-config.json          # Auto-detected project config
+├── .qa-context.json         # Auto-detected project config and session state
 ├── .qa-activity.log         # Session activity log
 ├── pr-reviews/
 ├── bug-reports/
@@ -140,7 +140,7 @@ qa-artifacts/
 └── flaky-diagnoses/
 ```
 
-### .qa-config.json Schema
+### .qa-context.json Schema
 
 | Field | Type | Description |
 |-------|------|-------------|
@@ -157,7 +157,7 @@ qa-artifacts/
 | `existingDocs` | object[] | Detected documentation files |
 | `existingTestDirs` | string[] | Detected test directories |
 
-**Note:** `projectRoot` is machine-specific — don't commit `.qa-config.json` to version control. The config is cached for 24 hours; delete it to force re-detection.
+**Note:** `projectRoot` is machine-specific — don't commit `.qa-context.json` to version control. Delete it to force re-detection.
 
 ---
 
@@ -167,7 +167,12 @@ The plugin ships a `.claude/settings.local.json` that pre-approves the following
 
 | Permission | Used by |
 |------------|---------|
-| `Bash(git:*)` | PR review skill — reads diff and log |
+| `Bash(git diff:*)` | PR review skill — reads PR diff |
+| `Bash(git log:*)` | PR review skill — reads commit history |
+| `Bash(git show:*)` | PR review skill — reads commit details |
+| `Bash(git branch:*)` | PR review skill — lists branches |
+| `Bash(git status:*)` | PR review skill — reads working tree state |
+| `Bash(git fetch:*)` | PR review skill — updates remote refs for comparison |
 | `Bash(node scripts/detect-project.js:*)` | SessionStart hook — project detection |
 | `Bash(node scripts/session-hook.js:*)` | Stop hook — session archiving and activity log |
 | `Bash(node scripts/state-manager.js:*)` | All skills — reading and writing QA state |
@@ -175,8 +180,10 @@ The plugin ships a `.claude/settings.local.json` that pre-approves the following
 You'll see these listed when you install the plugin and can revoke them at any time.
 
 **Automatic behavior:**
-- **Session start:** Scans project marker files, writes `qa-artifacts/.qa-config.json`. No network calls.
-- **Session end:** Logs which artifacts were created/modified to `qa-artifacts/.qa-activity.log`. No network calls.
+- **Session start:** Scans project marker files (languages, frameworks, test tools, CI/CD), reads the first 50 lines of `CLAUDE.md` if present (to detect project conventions), writes `qa-artifacts/.qa-context.json`. No network calls.
+- **Session end:** Promotes session findings to project state, archives the session, logs which artifacts were created/modified to `qa-artifacts/.qa-activity.log`. No network calls.
+
+**Output directory:** All artifacts are written to `qa-artifacts/` in your project root. This directory is runtime-generated and should be gitignored — add `qa-artifacts/` to your project's `.gitignore` if you don't want to commit QA artifacts.
 
 **To disable hooks** without removing the plugin:
 ```json
@@ -199,6 +206,8 @@ claude plugin add github:cyberwalk3r/qa-toolkit
 claude plugin add ./qa-toolkit
 ```
 
+The plugin ships with `.claude/settings.local.json` that pre-approves its own hook scripts and read-only git commands so everything works on first use without confirmation prompts. See the [Permissions & Side Effects](#permissions--side-effects) section for the full list.
+
 ---
 
 ## Works With Your Existing Setup
diff --git a/docs/index.html b/docs/index.html
index 997b844..9b28f88 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -3,89 +3,135 @@
 <head>
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
-  <title>QA Toolkit — QA that thinks for itself</title>
-  <meta name="description" content="A Claude Code plugin that auto-detects your stack and produces structured QA artifacts from plain English descriptions.">
+  <title>QA Toolkit — AI-powered QA slash commands for Claude Code</title>
+  <meta name="description" content="QA Toolkit is a free Claude Code plugin with 15 slash commands for test case generation, PR review, bug reports, regression planning, and exploratory testing. Auto-detects your stack. Zero config.">
+  <meta name="keywords" content="Claude Code plugin, QA automation, test case generation, PR review tool, bug report generator, regression testing, playwright, jest, pytest, quality assurance AI, software testing">
+  <meta name="author" content="cyberwalk3r">
+  <meta name="robots" content="index, follow">
+  <meta name="theme-color" content="#f59e0b">
+  <link rel="canonical" href="https://cyberwalk3r.github.io/qa-toolkit/">
 
   <!-- Favicon -->
   <link rel="icon" type="image/png" href="favicon.png">
 
   <!-- Open Graph -->
-  <meta property="og:title" content="QA Toolkit — QA that thinks for itself">
-  <meta property="og:description" content="15 slash commands. Auto-detects your stack. Produces structured QA artifacts from plain English descriptions.">
+  <meta property="og:title" content="QA Toolkit — AI-powered QA slash commands for Claude Code">
+  <meta property="og:description" content="15 slash commands for test cases, PR reviews, bug reports, regression plans, and more. Auto-detects your stack. Free Claude Code plugin.">
   <meta property="og:image" content="https://cyberwalk3r.github.io/qa-toolkit/og-image.png">
   <meta property="og:url" content="https://cyberwalk3r.github.io/qa-toolkit/">
   <meta property="og:type" content="website">
+  <meta property="og:site_name" content="QA Toolkit">
 
   <!-- Twitter Card -->
   <meta name="twitter:card" content="summary_large_image">
-  <meta name="twitter:title" content="QA Toolkit — QA that thinks for itself">
-  <meta name="twitter:description" content="15 slash commands. Auto-detects your stack. Produces structured QA artifacts from plain English descriptions.">
+  <meta name="twitter:title" content="QA Toolkit — AI-powered QA slash commands for Claude Code">
+  <meta name="twitter:description" content="15 slash commands for test cases, PR reviews, bug reports, regression plans, and more. Auto-detects your stack. Free Claude Code plugin.">
   <meta name="twitter:image" content="https://cyberwalk3r.github.io/qa-toolkit/og-image.png">
 
+  <!-- JSON-LD Structured Data -->
+  <script type="application/ld+json">
+  {
+    "@context": "https://schema.org",
+    "@type": "SoftwareApplication",
+    "name": "QA Toolkit",
+    "description": "A Claude Code plugin providing 15 QA slash commands for test case generation, PR review, bug reports, regression planning, exploratory testing, and more. Auto-detects your tech stack.",
+    "applicationCategory": "DeveloperApplication",
+    "operatingSystem": "Any",
+    "offers": {
+      "@type": "Offer",
+      "price": "0",
+      "priceCurrency": "USD"
+    },
+    "url": "https://cyberwalk3r.github.io/qa-toolkit/",
+    "downloadUrl": "https://github.com/cyberwalk3r/qa-toolkit",
+    "author": {
+      "@type": "Person",
+      "name": "cyberwalk3r",
+      "url": "https://github.com/cyberwalk3r"
+    },
+    "softwareVersion": "2.0.0",
+    "license": "https://opensource.org/licenses/MIT"
+  }
+  </script>
+
   <link rel="stylesheet" href="style.css">
 </head>
 <body>
 
 <nav>
   <span class="logo">/qa-toolkit</span>
-  <div>
+  <div class="nav-links">
     <a href="#how-it-works">How It Works</a>
     <a href="#commands">Commands</a>
     <a href="#agents">Agents</a>
     <a href="#install">Install</a>
-    <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank">GitHub</a>
+    <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank" rel="noopener noreferrer">GitHub</a>
   </div>
 </nav>
 
 <!-- Hero -->
 <section class="hero">
-  <h1>QA that <span>thinks for itself.</span></h1>
-  <p class="tagline">Describe what you're testing. Get a structured, stack-aware artifact — instantly. No templates. No formatting. No busywork.</p>
+  <div class="hero-glow" aria-hidden="true"></div>
+  <div class="hero-grid" aria-hidden="true"></div>
 
-  <div class="install-block">
-    <span class="prompt">$</span>
-    <span id="install-cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
-    <button class="copy-btn" onclick="copyInstall()">copy</button>
-  </div>
+  <div class="hero-content">
+    <div class="hero-badge">Claude Code Plugin &nbsp;·&nbsp; v2.0 &nbsp;·&nbsp; Free &amp; Open Source</div>
 
-  <div class="terminal" aria-label="Terminal demo">
-    <div class="terminal-bar">
-      <span class="dot red"></span>
-      <span class="dot yellow"></span>
-      <span class="dot green"></span>
-      <span class="terminal-title">Claude Code</span>
+    <h1>QA that <span class="shimmer">thinks for itself.</span></h1>
+    <p class="tagline">Describe what you're testing. Get a structured, stack-aware artifact — instantly. No templates. No formatting. No busywork.</p>
+
+    <div class="install-block">
+      <span class="prompt">$</span>
+      <span id="install-cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
+      <button class="copy-btn" onclick="copyInstall()" aria-label="Copy install command">copy</button>
+    </div>
+
+    <div class="stats-strip" aria-label="Quick stats">
+      <div class="stat">
+        <span class="stat-num">15</span>
+        <span class="stat-label">slash commands</span>
+      </div>
+      <div class="stat-divider" aria-hidden="true"></div>
+      <div class="stat">
+        <span class="stat-num">3</span>
+        <span class="stat-label">QA agents</span>
+      </div>
+      <div class="stat-divider" aria-hidden="true"></div>
+      <div class="stat">
+        <span class="stat-num">0</span>
+        <span class="stat-label">config required</span>
+      </div>
+      <div class="stat-divider" aria-hidden="true"></div>
+      <div class="stat">
+        <span class="stat-num">8+</span>
+        <span class="stat-label">tech stacks</span>
+      </div>
     </div>
-    <div class="terminal-body">
-      <div class="line-cmd typewriter">&gt; /qa-toolkit:bug-report The checkout page crashes with long addresses</div>
-      <br>
-      <div class="line-out" style="animation-delay:2.5s">Detecting project context...</div>
-      <div class="line-key">Stack: <span class="line-val">React · Node.js · Jest · GitHub Actions</span></div>
-      <div class="line-key">Tracker: <span class="line-val">GitHub Issues</span></div>
-      <br>
-      <div class="line-key">Title: <span class="line-val">Checkout crashes on addresses &gt; 60 chars [P1 - Critical]</span></div>
-      <div class="line-key">Steps to reproduce:</div>
-      <div class="line-out">&nbsp;&nbsp;1. Navigate to /checkout</div>
-      <div class="line-out">&nbsp;&nbsp;2. Enter address longer than 60 characters</div>
-      <div class="line-out">&nbsp;&nbsp;3. Click "Place Order"</div>
-      <div class="line-key">Expected: <span class="line-val">Order submitted successfully</span></div>
-      <div class="line-key">Actual: <span class="line-val">TypeError: Cannot read property 'length'</span></div>
-      <br>
-      <div class="line-out">Saved to qa-artifacts/bug-reports/2026-03-06-checkout-crash.md</div>
-      <span class="cursor"></span>
+
+    <div class="terminal" aria-label="Terminal demo">
+      <div class="terminal-bar" aria-hidden="true">
+        <span class="dot red"></span>
+        <span class="dot yellow"></span>
+        <span class="dot green"></span>
+        <span class="terminal-title">Claude Code</span>
+      </div>
+      <div class="terminal-body" id="terminal-body"></div>
     </div>
+
+    <p class="hero-hint">↓ scroll to see all 15 commands</p>
   </div>
 </section>
 
-<!-- What you stop doing -->
+<!-- Value props -->
 <div class="quotes">
   <div class="quotes-inner">
-    <div class="quote">
+    <div class="quote" data-animate>
       <p><strong>Bug reports in seconds.</strong> Describe what broke. Get back severity classification, steps to reproduce, expected vs actual, and environment fields — formatted for your tracker, not a generic template.</p>
     </div>
-    <div class="quote">
+    <div class="quote" data-animate>
       <p><strong>PR reviews that actually help.</strong> It reads the diff, maps which files touch which user flows, and writes a QA checklist grounded in what actually changed — not a checklist you've seen a hundred times.</p>
     </div>
-    <div class="quote">
+    <div class="quote" data-animate>
       <p><strong>Artifacts that fit your stack.</strong> It knows you're using Jest, not Pytest. It knows your CI runs on GitHub Actions. The output matches what your team already does, automatically.</p>
     </div>
   </div>
@@ -98,17 +144,17 @@ <h1>QA that <span>thinks for itself.</span></h1>
     <h2>Three steps. That's it.</h2>
     <p class="section-sub">Install once. Open any project. It scans your stack before you type your first command.</p>
     <div class="steps">
-      <div class="step">
+      <div class="step" data-animate>
         <div class="step-num">01</div>
         <h3>Install the plugin</h3>
         <p>One command. Works with any project — React, Django, .NET, Go, whatever you're building.</p>
       </div>
-      <div class="step">
+      <div class="step" data-animate>
         <div class="step-num">02</div>
         <h3>Open your project</h3>
         <p>On session start, QA Toolkit scans your stack and saves context. Knows your frameworks, test tools, and CI setup automatically.</p>
       </div>
-      <div class="step">
+      <div class="step" data-animate>
         <div class="step-num">03</div>
         <h3>Describe what you need</h3>
         <p>Type a slash command in plain English. Because it already knows your stack, the output uses the right format, the right frameworks, the right test runner.</p>
@@ -118,13 +164,13 @@ <h3>Describe what you need</h3>
 </section>
 
 <!-- Commands -->
-<section id="commands" style="background: var(--surface); border-top: 1px solid var(--border); border-bottom: 1px solid var(--border);">
+<section id="commands" class="commands-section">
   <div class="container">
     <div class="section-label">Commands</div>
     <h2>15 slash commands. Every QA workflow covered.</h2>
-    <p class="section-sub">Grouped by where they fit in your workflow. Each one produces a formatted artifact saved to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/</code>.</p>
+    <p class="section-sub">Grouped by where they fit in your workflow. Each one produces a formatted artifact saved to <code class="inline-code">qa-artifacts/</code>.</p>
 
-    <div class="phase-group">
+    <div class="phase-group" data-animate>
       <div class="phase-label">Authoring</div>
       <div class="command-grid">
         <div class="command-card"><code>/qa-toolkit:test-cases</code><p>Test cases from requirements — table, Gherkin, or checklist format</p></div>
@@ -135,7 +181,7 @@ <h2>15 slash commands. Every QA workflow covered.</h2>
       </div>
     </div>
 
-    <div class="phase-group">
+    <div class="phase-group" data-animate>
       <div class="phase-label">Review</div>
       <div class="command-grid">
         <div class="command-card"><code>/qa-toolkit:pr-review</code><p>Risk-flagged PR review with plain-English summary and QA checklist</p></div>
@@ -144,7 +190,7 @@ <h2>15 slash commands. Every QA workflow covered.</h2>
       </div>
     </div>
 
-    <div class="phase-group">
+    <div class="phase-group" data-animate>
       <div class="phase-label">Release</div>
       <div class="command-grid">
         <div class="command-card"><code>/qa-toolkit:regression-planner</code><p>Risk-based regression plan with time estimates</p></div>
@@ -153,7 +199,7 @@ <h2>15 slash commands. Every QA workflow covered.</h2>
       </div>
     </div>
 
-    <div class="phase-group">
+    <div class="phase-group" data-animate>
       <div class="phase-label">Analysis</div>
       <div class="command-grid">
         <div class="command-card"><code>/qa-toolkit:coverage-gap</code><p>Coverage gap analysis against requirements or test plan</p></div>
@@ -172,28 +218,31 @@ <h2>15 slash commands. Every QA workflow covered.</h2>
     <h2>When you need a conversation, not a document.</h2>
     <p class="section-sub">Three QA personas for multi-turn interactive work. They remember context across the conversation and adapt as you share more.</p>
     <div class="agent-grid">
-      <div class="agent-card">
+      <div class="agent-card" data-animate>
+        <div class="agent-icon" aria-hidden="true">◈</div>
         <h3>qa-reviewer</h3>
         <p class="role">The Code Reviewer's QA Brain</p>
         <p>Walk me through this PR. What should I actually be testing? Translates code changes into testing impact in plain English.</p>
       </div>
-      <div class="agent-card">
+      <div class="agent-card" data-animate>
+        <div class="agent-icon" aria-hidden="true">◈</div>
         <h3>qa-explorer</h3>
         <p class="role">The Edge Case Hunter</p>
         <p>Help me find what could break in this feature. Generates exploratory test charters and surfaces edge cases you wouldn't think to test.</p>
       </div>
-      <div class="agent-card">
+      <div class="agent-card" data-animate>
+        <div class="agent-icon" aria-hidden="true">◈</div>
         <h3>qa-lead</h3>
         <p class="role">The Release Decision Maker</p>
         <p>Let's decide if we're ready to ship. Makes release decisions, plans regression scope, and produces executive summaries.</p>
       </div>
     </div>
 
-    <div class="v2-callout">
+    <div class="v2-callout" data-animate>
       <h3>What's New in v2.0</h3>
       <ul>
         <li>State-aware skills — every command reads your detected project context</li>
-        <li>15 skills — 5 new in v2: test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis</li>
+        <li>15 skills — 5 new: test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis</li>
         <li>Redesigned agents — explicit tool boundaries, typed return contracts, persistent memory</li>
         <li>Multi-format output — artifacts in the format that actually fits the deliverable</li>
         <li>Shared references — consistent behavior across all skills</li>
@@ -208,28 +257,126 @@ <h3>What's New in v2.0</h3>
     <div class="section-label">Install</div>
     <h2>One command to install. Zero to configure.</h2>
     <p class="section-sub">Works with any project. Configurable if you want — but you don't have to.</p>
-    <div class="code-block"><span class="comment"># From GitHub</span>
+    <div class="code-block" data-animate><span class="comment"># From GitHub</span>
 <span class="cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
 
 <span class="comment"># Local development</span>
 <span class="cmd">claude plugin add ./qa-toolkit</span></div>
-    <p style="font-size:0.875rem;color:var(--text-muted)">On first session, auto-detection runs and saves your project context to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/.qa-config.json</code>. Every subsequent command reads from it.</p>
+    <p class="install-note">On first session, auto-detection runs and saves your project context to <code class="inline-code">qa-artifacts/.qa-context.json</code>. Every subsequent command reads from it.</p>
   </div>
 </section>
 
 <footer>
-  <p>MIT License &nbsp;·&nbsp; <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank">GitHub</a> &nbsp;·&nbsp; Built for Claude Code</p>
+  <p>MIT License &nbsp;·&nbsp; <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank" rel="noopener noreferrer">GitHub</a> &nbsp;·&nbsp; Built for Claude Code</p>
 </footer>
 
 <script>
+// ---- Copy install command ----
 function copyInstall() {
   const text = document.getElementById('install-cmd').textContent;
   navigator.clipboard.writeText(text).then(() => {
     const btn = document.querySelector('.copy-btn');
     btn.textContent = 'copied!';
-    setTimeout(() => btn.textContent = 'copy', 2000);
+    btn.classList.add('copied');
+    setTimeout(() => { btn.textContent = 'copy'; btn.classList.remove('copied'); }, 2000);
   });
 }
+
+// ---- Terminal cycling demos ----
+const demos = [
+  {
+    cmd: '> /qa-toolkit:bug-report The checkout page crashes with long addresses',
+    lines: [
+      { cls: 'line-out', text: 'Detecting project context...' },
+      { cls: 'line-key', text: 'Stack: <span class="line-val">React · Node.js · Jest · GitHub Actions</span>' },
+      { cls: 'line-key', text: 'Tracker: <span class="line-val">GitHub Issues</span>' },
+      { cls: 'line-spacer', text: '' },
+      { cls: 'line-key', text: 'Title: <span class="line-val">Checkout crashes on addresses &gt; 60 chars [P1]</span>' },
+      { cls: 'line-key', text: 'Actual: <span class="line-val">TypeError: Cannot read property \'length\'</span>' },
+      { cls: 'line-out', text: 'Saved → qa-artifacts/bug-reports/2026-03-06-checkout-crash.md' },
+    ]
+  },
+  {
+    cmd: '> /qa-toolkit:pr-review feature/new-auth-flow',
+    lines: [
+      { cls: 'line-out', text: 'Reading diff · 12 files changed...' },
+      { cls: 'line-key', text: 'Risk Level: <span class="line-val line-amber">HIGH — auth flow changed</span>' },
+      { cls: 'line-key', text: 'Affected flows: <span class="line-val">login · session · password-reset</span>' },
+      { cls: 'line-spacer', text: '' },
+      { cls: 'line-key', text: 'QA Checklist: <span class="line-val">7 items generated</span>' },
+      { cls: 'line-out', text: '  ✓ Verify session expiry after token rotation' },
+      { cls: 'line-out', text: 'Saved → qa-artifacts/pr-reviews/pr-review-2026-03-06-auth-flow.md' },
+    ]
+  },
+  {
+    cmd: '> /qa-toolkit:test-cases User registration with email verification',
+    lines: [
+      { cls: 'line-out', text: 'Reading project context...' },
+      { cls: 'line-key', text: 'Framework: <span class="line-val">Playwright detected — using BDD format</span>' },
+      { cls: 'line-spacer', text: '' },
+      { cls: 'line-key', text: 'TC-REG-01 <span class="line-val">P0 · Happy path registration</span>' },
+      { cls: 'line-key', text: 'TC-REG-02 <span class="line-val">P0 · Duplicate email rejected</span>' },
+      { cls: 'line-key', text: 'TC-REG-03 <span class="line-val">P1 · Verification link expires</span>' },
+      { cls: 'line-out', text: 'Saved → qa-artifacts/test-cases/test-cases-2026-03-06-registration.md' },
+    ]
+  }
+];
+
+let currentDemo = 0;
+const terminalBody = document.getElementById('terminal-body');
+
+function renderDemo(idx, animate) {
+  const demo = demos[idx];
+  terminalBody.innerHTML = '';
+
+  const cmdDiv = document.createElement('div');
+  cmdDiv.className = 'line-cmd' + (animate ? ' typewriter' : '');
+  cmdDiv.textContent = demo.cmd;
+  terminalBody.appendChild(cmdDiv);
+
+  const revealDelay = animate ? 2600 : 0;
+
+  demo.lines.forEach((line, i) => {
+    const div = document.createElement('div');
+    div.className = line.cls || 'line-spacer';
+    div.innerHTML = line.text;
+    div.style.opacity = '0';
+    div.style.transition = 'opacity 0.35s ease';
+    terminalBody.appendChild(div);
+    setTimeout(() => { div.style.opacity = '1'; }, revealDelay + i * 200);
+  });
+
+  const cursor = document.createElement('span');
+  cursor.className = 'cursor';
+  terminalBody.appendChild(cursor);
+}
+
+function cycleDemo() {
+  terminalBody.style.opacity = '0';
+  setTimeout(() => {
+    currentDemo = (currentDemo + 1) % demos.length;
+    terminalBody.style.opacity = '1';
+    renderDemo(currentDemo, true);
+  }, 400);
+}
+
+renderDemo(0, true);
+setInterval(cycleDemo, 7000);
+
+// ---- Scroll reveal ----
+const observer = new IntersectionObserver((entries) => {
+  entries.forEach(entry => {
+    if (entry.isIntersecting) {
+      entry.target.classList.add('visible');
+      observer.unobserve(entry.target);
+    }
+  });
+}, { threshold: 0.08 });
+
+document.querySelectorAll('[data-animate]').forEach((el, i) => {
+  el.style.transitionDelay = (i % 4 * 0.1) + 's';
+  observer.observe(el);
+});
 </script>
 
 </body>
diff --git a/docs/plans/2026-02-23-plugin-improvements-design.md b/docs/plans/2026-02-23-plugin-improvements-design.md
deleted file mode 100644
index f75b640..0000000
--- a/docs/plans/2026-02-23-plugin-improvements-design.md
+++ /dev/null
@@ -1,58 +0,0 @@
-# QA Toolkit Plugin Improvements Design
-
-Date: 2026-02-23
-
-## Summary
-
-Fix 10 identified weaknesses in the qa-toolkit v1.0.0 plugin: 2 critical bugs, 4 significant issues, and 4 design gaps.
-
-## Fixes
-
-### 1. Hook event: TaskCompleted → Stop (Critical)
-
-`TaskCompleted` is not a valid Claude Code hook event. Change `hooks.json` to use `Stop`. No changes to `save-artifact.js` logic.
-
-### 2. Remove dead code in detect-project.js (Critical)
-
-Delete unused `detectors.frameworks` (lines 53-66) and `detectors.testFrameworks` (lines 67-86) arrays. Actual detection uses `frameworkChecks` and `testChecks` arrays.
-
-### 3. Lazy directory creation (Significant)
-
-Stop creating 9 empty subdirectories on SessionStart. Only create `qa-artifacts/` root for the config file. Each skill creates its subdirectory on first artifact save.
-
-### 4. Fix pyproject.toml parsing (Significant)
-
-Replace crude regex with section-aware parsing. Only extract dependency names from `[project.dependencies]`, `[project.optional-dependencies]`, and `[tool.poetry.dependencies]` sections.
-
-### 5. Recursive glob for .csproj/.sln (Significant)
-
-Replace root-only `readdirSync` with recursive walk (depth limit 3) to catch `.csproj` files in subdirectories.
-
-### 6. Validate browser arg in run-test.js (Significant)
-
-Whitelist check against `['chromium', 'firefox', 'webkit']`. Exit with clear error if invalid.
-
-### 7. Skill cross-referencing (Design)
-
-Add conditional "Suggested Next Steps" section to each skill's output template:
-- pr-review → regression-planner (if risk Medium+), test-cases
-- bug-report → test-cases
-- test-cases → test-data, e2e-test
-- api-test → test-data
-- e2e-test → accessibility
-- regression-planner → release-readiness
-- accessibility → bug-report
-- release-readiness → regression-planner (if score low)
-- setup → pr-review, test-cases
-
-### 8. Add CLAUDE.md (Design)
-
-Create repo-level CLAUDE.md covering plugin structure, conventions, testing instructions.
-
-### 9. Configurable output directory (Design)
-
-Add `"outputDir": "qa-artifacts"` to `settings.json`. Scripts read this setting. Skills reference configured directory from `.qa-config.json`.
-
-### 10. Clarify agent/skill boundary (Design)
-
-Update agent descriptions: conversational, multi-turn QA work. Update skill descriptions: one-shot structured output. Add "Agents vs Skills" section to README.
diff --git a/docs/plans/2026-02-23-plugin-improvements.md b/docs/plans/2026-02-23-plugin-improvements.md
deleted file mode 100644
index 3b4acf6..0000000
--- a/docs/plans/2026-02-23-plugin-improvements.md
+++ /dev/null
@@ -1,667 +0,0 @@
-# QA Toolkit v1.1.0 Improvements — Implementation Plan
-
-**Goal:** Fix 10 weaknesses in qa-toolkit v1.0.0 — 2 critical bugs, 4 significant issues, 4 design gaps.
-
-**Architecture:** All changes are to existing files (scripts, hooks config, skill markdown, agent markdown, README). No new runtime dependencies. One new file: CLAUDE.md.
-
-**Tech Stack:** Node.js scripts, Markdown skills/agents, JSON config.
-
-**Design doc:** `docs/plans/2026-02-23-plugin-improvements-design.md`
-
----
-
-### Task 1: Fix broken hook event — TaskCompleted → Stop
-
-**Files:**
-- Modify: `hooks/hooks.json`
-
-**Step 1: Fix the hook event name**
-
-In `hooks/hooks.json`, replace `"TaskCompleted"` with `"Stop"`:
-
-```json
-{
-    "hooks": {
-        "SessionStart": [
-            {
-                "hooks": [
-                    {
-                        "type": "command",
-                        "command": "node \"${CLAUDE_PLUGIN_ROOT}/scripts/detect-project.js\""
-                    }
-                ]
-            }
-        ],
-        "Stop": [
-            {
-                "hooks": [
-                    {
-                        "type": "command",
-                        "command": "node \"${CLAUDE_PLUGIN_ROOT}/scripts/save-artifact.js\""
-                    }
-                ]
-            }
-        ]
-    }
-}
-```
-
-**Step 2: Commit**
-
-```bash
-git add hooks/hooks.json
-git commit -s -m "fix: replace invalid TaskCompleted hook with Stop event"
-```
-
----
-
-### Task 2: Remove dead code in detect-project.js
-
-**Files:**
-- Modify: `scripts/detect-project.js`
-
-**Step 1: Delete the unused `detectors.frameworks` and `detectors.testFrameworks` arrays**
-
-In `scripts/detect-project.js`, the `detectors` object (lines 32-108) has four properties: `languages`, `frameworks`, `testFrameworks`, `cicd`, `packageManagers`. Only `languages`, `cicd`, and `packageManagers` are iterated. `frameworks` and `testFrameworks` are dead — the real work is done by `frameworkChecks` (line 230) and `testChecks` (line 252).
-
-Delete the `frameworks` array (lines 53-66) and the `testFrameworks` array (lines 67-86) from the `detectors` object. Keep `languages`, `cicd`, and `packageManagers`.
-
-The resulting `detectors` object should be:
-
-```javascript
-const detectors = {
-    languages: [
-        { marker: 'package.json', lang: 'JavaScript/TypeScript' },
-        { marker: 'tsconfig.json', lang: 'TypeScript' },
-        { marker: 'requirements.txt', lang: 'Python' },
-        { marker: 'pyproject.toml', lang: 'Python' },
-        { marker: 'setup.py', lang: 'Python' },
-        { marker: 'Pipfile', lang: 'Python' },
-        { marker: 'go.mod', lang: 'Go' },
-        { marker: 'Cargo.toml', lang: 'Rust' },
-        { marker: 'pom.xml', lang: 'Java' },
-        { marker: 'build.gradle', lang: 'Java/Kotlin' },
-        { marker: 'build.gradle.kts', lang: 'Kotlin' },
-        { marker: '*.csproj', lang: 'C#/.NET', glob: true },
-        { marker: '*.sln', lang: 'C#/.NET', glob: true },
-        { marker: 'Gemfile', lang: 'Ruby' },
-        { marker: 'composer.json', lang: 'PHP' },
-        { marker: 'mix.exs', lang: 'Elixir' },
-        { marker: 'Package.swift', lang: 'Swift' },
-        { marker: 'pubspec.yaml', lang: 'Dart/Flutter' },
-    ],
-    cicd: [
-        { marker: '.github/workflows', name: 'GitHub Actions', isDir: true },
-        { marker: '.gitlab-ci.yml', name: 'GitLab CI' },
-        { marker: 'Jenkinsfile', name: 'Jenkins' },
-        { marker: 'azure-pipelines.yml', name: 'Azure Pipelines' },
-        { marker: '.circleci', name: 'CircleCI', isDir: true },
-        { marker: 'bitbucket-pipelines.yml', name: 'Bitbucket Pipelines' },
-        { marker: '.travis.yml', name: 'Travis CI' },
-        { marker: 'Dockerfile', name: 'Docker' },
-        { marker: 'docker-compose.yml', name: 'Docker Compose' },
-        { marker: 'docker-compose.yaml', name: 'Docker Compose' },
-    ],
-    packageManagers: [
-        { marker: 'pnpm-lock.yaml', name: 'pnpm' },
-        { marker: 'yarn.lock', name: 'Yarn' },
-        { marker: 'package-lock.json', name: 'npm' },
-        { marker: 'bun.lockb', name: 'Bun' },
-        { marker: 'Pipfile.lock', name: 'Pipenv' },
-        { marker: 'poetry.lock', name: 'Poetry' },
-        { marker: 'uv.lock', name: 'uv' },
-    ],
-};
-```
-
-**Step 2: Commit**
-
-```bash
-git add scripts/detect-project.js
-git commit -s -m "fix: remove dead detectors.frameworks and detectors.testFrameworks arrays"
-```
-
----
-
-### Task 3: Lazy directory creation
-
-**Files:**
-- Modify: `scripts/detect-project.js`
-
-**Step 1: Remove subdirectory creation from detect-project.js**
-
-In `scripts/detect-project.js`, find the block that creates artifact subdirectories (around lines 307-316 after Task 2 edits):
-
-```javascript
-// Create artifact subdirectories
-const artifactDirs = [
-    'pr-reviews', 'bug-reports', 'test-cases', 'api-tests',
-    'regression-plans', 'test-data', 'a11y-audits',
-    'release-assessments', 'e2e-tests'
-];
-for (const dir of artifactDirs) {
-    const dirPath = path.join(configDir, dir);
-    if (!fs.existsSync(dirPath)) fs.mkdirSync(dirPath, { recursive: true });
-}
-```
-
-Delete this entire block. Keep the `qa-artifacts/` root creation (line ~302) because the config JSON goes there.
-
-**Step 2: Commit**
-
-```bash
-git add scripts/detect-project.js
-git commit -s -m "fix: stop eagerly creating artifact subdirectories on session start"
-```
-
----
-
-### Task 4: Fix pyproject.toml parsing
-
-**Files:**
-- Modify: `scripts/detect-project.js`
-
-**Step 1: Replace the crude regex with section-aware parsing**
-
-Find the pyproject.toml parsing block (around lines 224-228 after previous edits):
-
-```javascript
-// Check pyproject.toml dependencies too
-try {
-    const pyproject = fs.readFileSync(path.join(projectRoot, 'pyproject.toml'), 'utf8').toLowerCase();
-    const depMatches = pyproject.match(/["']([a-z0-9_-]+)["']/g) || [];
-    pyDeps.push(...depMatches.map(m => m.replace(/["']/g, '')));
-} catch { }
-```
-
-Replace with:
-
-```javascript
-// Check pyproject.toml dependencies — section-aware parsing
-try {
-    const pyproject = fs.readFileSync(path.join(projectRoot, 'pyproject.toml'), 'utf8');
-    const lines = pyproject.split('\n');
-    let inDepSection = false;
-    const depSections = [
-        '[project.dependencies]',
-        '[project.optional-dependencies',
-        '[tool.poetry.dependencies]',
-        '[tool.poetry.dev-dependencies]',
-        '[tool.poetry.group.',
-    ];
-    for (const line of lines) {
-        const trimmed = line.trim().toLowerCase();
-        if (trimmed.startsWith('[')) {
-            inDepSection = depSections.some(s => trimmed.startsWith(s));
-            continue;
-        }
-        if (inDepSection && trimmed && !trimmed.startsWith('#')) {
-            // Match dependency name: "package-name", 'package-name', or bare package-name (before = or >)
-            const match = trimmed.match(/^["']?([a-z0-9_-]+)/);
-            if (match) pyDeps.push(match[1]);
-        }
-    }
-} catch { }
-```
-
-**Step 2: Commit**
-
-```bash
-git add scripts/detect-project.js
-git commit -s -m "fix: parse pyproject.toml dependencies by section instead of regex-matching all strings"
-```
-
----
-
-### Task 5: Recursive glob for .csproj/.sln
-
-**Files:**
-- Modify: `scripts/detect-project.js`
-
-**Step 1: Replace `hasGlobMatch` with recursive version**
-
-Find the current `hasGlobMatch` function (around line 119):
-
-```javascript
-function hasGlobMatch(pattern) {
-    try {
-        const ext = pattern.replace('*', '');
-        const files = fs.readdirSync(projectRoot);
-        return files.some(f => f.endsWith(ext));
-    } catch { return false; }
-}
-```
-
-Replace with:
-
-```javascript
-function hasGlobMatch(pattern) {
-    const ext = pattern.replace('*', '');
-    function search(dir, depth) {
-        if (depth > 3) return false;
-        try {
-            const entries = fs.readdirSync(dir, { withFileTypes: true });
-            for (const entry of entries) {
-                if (entry.name.startsWith('.') || entry.name === 'node_modules') continue;
-                if (entry.isFile() && entry.name.endsWith(ext)) return true;
-                if (entry.isDirectory() && search(path.join(dir, entry.name), depth + 1)) return true;
-            }
-        } catch { /* skip unreadable dirs */ }
-        return false;
-    }
-    return search(projectRoot, 0);
-}
-```
-
-**Step 2: Commit**
-
-```bash
-git add scripts/detect-project.js
-git commit -s -m "fix: recursive glob matching (depth 3) for .csproj/.sln detection"
-```
-
----
-
-### Task 6: Validate browser arg in run-test.js
-
-**Files:**
-- Modify: `skills/e2e-test/scripts/run-test.js`
-
-**Step 1: Add browser validation after flag parsing**
-
-Find the browser parsing line (around line 55):
-
-```javascript
-const browser = browserIdx >= 0 ? args[browserIdx + 1] : 'chromium';
-```
-
-Add validation immediately after:
-
-```javascript
-const validBrowsers = ['chromium', 'firefox', 'webkit'];
-if (!validBrowsers.includes(browser)) {
-    console.error(`Error: Invalid browser "${browser}". Must be one of: ${validBrowsers.join(', ')}`);
-    process.exit(1);
-}
-```
-
-**Step 2: Commit**
-
-```bash
-git add skills/e2e-test/scripts/run-test.js
-git commit -s -m "fix: validate --browser arg against chromium/firefox/webkit whitelist"
-```
-
----
-
-### Task 7: Add configurable output directory
-
-**Files:**
-- Modify: `settings.json`
-- Modify: `scripts/detect-project.js`
-- Modify: `scripts/save-artifact.js`
-
-**Step 1: Add outputDir to settings.json**
-
-Replace the full content of `settings.json`:
-
-```json
-{
-    "agent": "qa-reviewer",
-    "outputDir": "qa-artifacts"
-}
-```
-
-**Step 2: Update detect-project.js to read outputDir from settings**
-
-At the top of `scripts/detect-project.js`, after the existing `const projectRoot` line, find:
-
-```javascript
-const projectRoot = process.cwd();
-const configDir = path.join(projectRoot, 'qa-artifacts');
-```
-
-Replace with:
-
-```javascript
-const projectRoot = process.cwd();
-
-// Read output directory from plugin settings
-let outputDir = 'qa-artifacts';
-try {
-    const pluginRoot = process.env.CLAUDE_PLUGIN_ROOT || path.join(__dirname, '..');
-    const settings = JSON.parse(fs.readFileSync(path.join(pluginRoot, 'settings.json'), 'utf8'));
-    if (settings.outputDir) outputDir = settings.outputDir;
-} catch { /* use default */ }
-
-const configDir = path.join(projectRoot, outputDir);
-```
-
-Also update the `outputContext` function — find the line:
-
-```javascript
-    lines.push(`Config: qa-artifacts/.qa-config.json`);
-```
-
-Replace with:
-
-```javascript
-    lines.push(`Config: ${outputDir}/.qa-config.json`);
-    lines.push(`Output Directory: ${outputDir}/`);
-```
-
-Also in the config object, add the outputDir so skills can read it:
-
-Find:
-
-```javascript
-const config = {
-    detectedAt: new Date().toISOString(),
-    projectRoot,
-```
-
-Replace with:
-
-```javascript
-const config = {
-    detectedAt: new Date().toISOString(),
-    projectRoot,
-    outputDir,
-```
-
-**Step 3: Update save-artifact.js to read outputDir**
-
-In `scripts/save-artifact.js`, find:
-
-```javascript
-const projectRoot = process.cwd();
-const configDir = path.join(projectRoot, 'qa-artifacts');
-```
-
-Replace with:
-
-```javascript
-const projectRoot = process.cwd();
-
-// Read output directory from config or default
-let outputDir = 'qa-artifacts';
-try {
-    const configPath = path.join(projectRoot, 'qa-artifacts', '.qa-config.json');
-    const config = JSON.parse(fs.readFileSync(configPath, 'utf8'));
-    if (config.outputDir) outputDir = config.outputDir;
-} catch { /* use default */ }
-
-const configDir = path.join(projectRoot, outputDir);
-```
-
-**Step 4: Commit**
-
-```bash
-git add settings.json scripts/detect-project.js scripts/save-artifact.js
-git commit -s -m "feat: configurable output directory via settings.json outputDir"
-```
-
----
-
-### Task 8: Add skill cross-referencing
-
-**Files:**
-- Modify: `skills/pr-review/SKILL.md`
-- Modify: `skills/bug-report/SKILL.md`
-- Modify: `skills/test-cases/SKILL.md`
-- Modify: `skills/api-test/SKILL.md`
-- Modify: `skills/e2e-test/SKILL.md`
-- Modify: `skills/regression-planner/SKILL.md`
-- Modify: `skills/accessibility/SKILL.md`
-- Modify: `skills/release-readiness/SKILL.md`
-- Modify: `skills/test-data/SKILL.md`
-- Modify: `skills/setup/SKILL.md`
-
-**Step 1: Add "Suggested Next Steps" section to each skill**
-
-Append to the end of each SKILL.md (before any existing final section or at the very end):
-
-**`skills/pr-review/SKILL.md`** — append:
-
-```markdown
-## Suggested Next Steps
-After generating the review, suggest based on results:
-- If Risk Level is **Medium or higher**: "Consider running `/qa-toolkit:regression-planner` to map the impact radius and plan targeted regression testing."
-- Always: "Generate test cases for the changed functionality with `/qa-toolkit:test-cases`."
-```
-
-**`skills/bug-report/SKILL.md`** — append:
-
-```markdown
-## Suggested Next Steps
-After generating the bug report, suggest:
-- "Create test cases to cover this bug scenario and prevent regression with `/qa-toolkit:test-cases`."
-```
-
-**`skills/test-cases/SKILL.md`** — append:
-
-```markdown
-## Suggested Next Steps
-After generating test cases, suggest:
-- "Generate synthetic test data for these scenarios with `/qa-toolkit:test-data`."
-- "Automate the P0/P1 cases as Playwright E2E tests with `/qa-toolkit:e2e-test`."
-```
-
-**`skills/api-test/SKILL.md`** — append:
-
-```markdown
-## Suggested Next Steps
-After generating API tests, suggest:
-- "Generate realistic test payloads with `/qa-toolkit:test-data`."
-```
-
-**`skills/e2e-test/SKILL.md`** — append:
-
-```markdown
-## Suggested Next Steps
-After generating E2E tests, suggest:
-- "Run an accessibility audit on the same pages with `/qa-toolkit:accessibility`."
-```
-
-**`skills/regression-planner/SKILL.md`** — append:
-
-```markdown
-## Suggested Next Steps
-After generating the regression plan, suggest:
-- "When regression testing is complete, assess release readiness with `/qa-toolkit:release-readiness`."
-```
-
-**`skills/accessibility/SKILL.md`** — append:
-
-```markdown
-## Suggested Next Steps
-After generating the audit, suggest:
-- "File each critical/major finding as a bug report with `/qa-toolkit:bug-report`."
-```
-
-**`skills/release-readiness/SKILL.md`** — append:
-
-```markdown
-## Suggested Next Steps
-After generating the assessment, suggest based on results:
-- If verdict is **NO-GO** or Regression Score is below 70: "Plan targeted regression testing with `/qa-toolkit:regression-planner`."
-```
-
-**`skills/test-data/SKILL.md`** — no cross-reference needed, skip.
-
-**`skills/setup/SKILL.md`** — append:
-
-```markdown
-## Suggested Next Steps
-After setup is complete, suggest:
-- "Start with a PR review (`/qa-toolkit:pr-review`) or generate test cases from your requirements (`/qa-toolkit:test-cases`)."
-```
-
-**Step 2: Commit**
-
-```bash
-git add skills/*/SKILL.md
-git commit -s -m "feat: add contextual Suggested Next Steps cross-references to all skills"
-```
-
----
-
-### Task 9: Add CLAUDE.md for the plugin repo
-
-**Files:**
-- Create: `CLAUDE.md`
-
-**Step 1: Create CLAUDE.md at repo root**
-
-```markdown
-# QA Toolkit — Development Guide
-
-## What This Is
-
-A Claude Code plugin providing 10 QA skills (slash commands) and 3 agent personas. Auto-detects project tech stack and produces structured QA artifacts.
-
-## Project Structure
-
-```
-.claude-plugin/plugin.json  — Plugin manifest (name, version, metadata)
-settings.json               — Default agent and plugin settings
-hooks/hooks.json            — Event hooks (SessionStart, Stop)
-scripts/detect-project.js   — Auto-detection script (runs on SessionStart)
-scripts/save-artifact.js    — Activity logger (runs on Stop)
-agents/<name>.md            — Agent persona definitions (conversational, multi-turn)
-skills/<name>/SKILL.md      — Skill definitions (one-shot structured output)
-skills/<name>/references/   — Domain-specific knowledge files
-skills/<name>/scripts/      — Helper scripts (e.g., Playwright runner)
-```
-
-## Conventions
-
-### Skills
-- `SKILL.md` must have YAML frontmatter with `name` and `description` (one sentence)
-- Skills produce one-shot structured output and save to the configured output directory
-- Each skill reads `<outputDir>/.qa-config.json` for project context
-- Reference files contain non-obvious, specific guidance — not generic info
-- Include a "Suggested Next Steps" section with conditional cross-references to related skills
-
-### Agents
-- Agent `.md` files have YAML frontmatter with `name` and `description`
-- Agents define personas for conversational, multi-turn QA work (ongoing review sessions, exploratory discussions)
-- Agents define tone and expertise, not workflows — workflows belong in skills
-
-### Scripts
-- All scripts are Node.js with no external dependencies (stdlib only)
-- Scripts read `settings.json` for plugin-level config (e.g., `outputDir`)
-- The output directory defaults to `qa-artifacts/` and is configurable via `settings.json`
-
-## Testing Locally
-
-```bash
-claude plugin add ./qa-toolkit
-claude --debug --plugin-dir ./qa-toolkit
-```
-
-## Output Directory
-
-All generated artifacts save to `<outputDir>/` (default: `qa-artifacts/`) in the target project. Subdirectories are created lazily on first artifact save, not eagerly on session start.
-```
-
-**Step 2: Commit**
-
-```bash
-git add CLAUDE.md
-git commit -s -m "docs: add CLAUDE.md with plugin development conventions"
-```
-
----
-
-### Task 10: Clarify agent/skill boundary
-
-**Files:**
-- Modify: `agents/qa-reviewer.md`
-- Modify: `agents/qa-explorer.md`
-- Modify: `agents/qa-lead.md`
-- Modify: `README.md`
-
-**Step 1: Update agent frontmatter descriptions to clarify they're for multi-turn work**
-
-**`agents/qa-reviewer.md`** — change the frontmatter `description` to:
-
-```yaml
-description: Conversational QA reviewer for multi-turn PR review sessions — translates code changes into testing impact through interactive discussion
-```
-
-**`agents/qa-explorer.md`** — change the frontmatter `description` to:
-
-```yaml
-description: Conversational exploratory testing persona for multi-turn sessions — generates edge cases and test scenarios through interactive discussion
-```
-
-**`agents/qa-lead.md`** — change the frontmatter `description` to:
-
-```yaml
-description: Conversational QA lead for multi-turn strategic sessions — release decisions, quality metrics, and test planning through interactive discussion
-```
-
-**Step 2: Add "Agents vs Skills" section to README.md**
-
-After the "## Agents" section and before "## Install", add:
-
-```markdown
-## Agents vs Skills
-
-**Skills** (`/qa-toolkit:*`) produce **one-shot structured output** — you describe what you need, you get a formatted artifact saved to disk. Use these for specific deliverables: a PR review, a bug report, a test plan.
-
-**Agents** are **conversational personas** for multi-turn interactive work. Use these when you want an ongoing QA discussion: "walk me through this PR," "help me explore edge cases for this feature," "let's assess whether we're ready to release." Agents remember context across the conversation and adapt their guidance as you share more information.
-
-| Need | Use |
-|------|-----|
-| Structured PR review document | `/qa-toolkit:pr-review` (skill) |
-| Interactive PR walkthrough with Q&A | `qa-reviewer` (agent) |
-| Bug report from a description | `/qa-toolkit:bug-report` (skill) |
-| Brainstorm what could break in a feature | `qa-explorer` (agent) |
-| Go/no-go release document | `/qa-toolkit:release-readiness` (skill) |
-| Strategic release planning discussion | `qa-lead` (agent) |
-```
-
-**Step 3: Commit**
-
-```bash
-git add agents/*.md README.md
-git commit -s -m "docs: clarify agent vs skill boundary — agents are conversational, skills are one-shot"
-```
-
----
-
-### Task 11: Update CHANGELOG.md
-
-**Files:**
-- Modify: `CHANGELOG.md`
-
-**Step 1: Add v1.1.0 entry**
-
-Prepend to `CHANGELOG.md` after the `# Changelog` header:
-
-```markdown
-## [1.1.0] — 2026-02-23
-
-### Fixed
-- Replace invalid `TaskCompleted` hook event with `Stop` — activity logging now works
-- Remove dead `detectors.frameworks` and `detectors.testFrameworks` arrays in detect-project.js
-- Stop eagerly creating 9 empty artifact subdirectories on session start
-- Parse pyproject.toml dependencies by TOML section instead of regex-matching all quoted strings
-- Recursive glob matching (depth 3) for .csproj/.sln detection in subdirectories
-- Validate `--browser` argument in run-test.js against chromium/firefox/webkit
-
-### Added
-- Configurable output directory via `settings.json` `outputDir` field
-- Contextual "Suggested Next Steps" cross-references in all skills
-- CLAUDE.md with plugin development conventions
-- "Agents vs Skills" section in README clarifying when to use each
-```
-
-**Step 2: Commit**
-
-```bash
-git add CHANGELOG.md
-git commit -s -m "docs: add v1.1.0 changelog entry"
-```
diff --git a/docs/plans/2026-03-06-docs-github-pages-design.md b/docs/plans/2026-03-06-docs-github-pages-design.md
deleted file mode 100644
index c5b7887..0000000
--- a/docs/plans/2026-03-06-docs-github-pages-design.md
+++ /dev/null
@@ -1,131 +0,0 @@
-# Design: Docs Rewrite + GitHub Pages
-
-**Date:** 2026-03-06
-**Status:** Approved
-
-## Problem
-
-QA Toolkit v2.0 shipped 9 phases of work — 5 new skills, redesigned agents, a state-aware system, shared references, and multi-format output — but the docs still reflect v1.1.0. The README lists 10 commands (v2 has 15). The CHANGELOG has no v2.0 entry. CONTRIBUTING.md doesn't mention any v2 conventions. There is no public-facing site.
-
-## Goal
-
-1. Rewrite README, CHANGELOG, and CONTRIBUTING with a new narrative that appeals to both QA professionals and developers.
-2. Launch a GitHub Pages site (`docs/index.html`) that serves as a captivating landing page and full reference — no build step required.
-
-## Approach: README-First (Single Source of Truth)
-
-Fix the README as the authoritative source. The GitHub Pages site renders that content with a dramatically improved visual design. Update one, the other stays current in spirit.
-
----
-
-## Doc Fixes
-
-### README.md — Full Rewrite
-
-**Narrative shift:** From "a plugin that produces QA artifacts" to "QA that thinks for itself."
-
-**Structure:**
-```
-[centered header + one-liner + badges]
-[quick install one-liner]
-[2–3 punchy social-proof quotes]
-[Why I Built This — short, honest, personal]
-[How It Works — 3-step flow with terminal example]
-[Commands — grouped by phase: Authoring / Review / Release / Analysis]
-[Agents — "when you need a conversation, not a document"]
-[What's New in v2.0 — state-awareness, 15 skills, redesigned agents]
-[Output — artifact directory tree]
-[Permissions & Side Effects]
-[Install options]
-[License]
-```
-
-**Tone:** Direct, slightly irreverent, anti-enterprise. Speaks to both the developer who also does QA and the dedicated QA lead.
-
-**Key additions vs current:**
-- All 15 skills (current README lists 10 — missing test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis)
-- Commands grouped by workflow phase instead of flat table
-- v2.0 state-aware system explained in plain language
-- Badges: install, GitHub stars, license, CI status
-- Personal founder voice in "Why I Built This"
-
-### CHANGELOG.md — Add v2.0 Entry
-
-Written as a narrative summary (not just bullets) that tells the story of what changed and why. Covers:
-- State-aware skill system
-- 5 new skills (test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis)
-- 3 redesigned skills (e2e-test, bug-report, api-test, regression-planner, pr-review)
-- Redesigned agents (tool restrictions, return contracts, persistent memory)
-- Shared references system (`skills/shared/references/`)
-- Multi-format output
-- Test suite updates
-
-### CONTRIBUTING.md — Rewrite
-
-Friendlier tone, structured as "here's how it all fits together." Adds:
-- State-aware skill pattern: read `.qa-config.json` → adapt output → write findings
-- Shared references in `skills/shared/references/` and when to use them
-- v2 agent conventions: tool restrictions, return contracts, persistent memory
-- Updated skill creation steps reflecting v2 conventions
-
----
-
-## GitHub Pages Site
-
-### Tech Stack
-
-Pure HTML/CSS/JS in `docs/` folder. GitHub Pages serves from `docs/` on `main` branch — no build step, no CI, no config files needed.
-
-### Visual Direction
-
-- Dark background (`#0d0d0d`), monospace font for terminal aesthetic
-- Accent color: electric amber (`#f59e0b`) — warm, readable, distinct from GSD green
-- Hero: full-viewport, animated CSS typewriter showing a `/qa-toolkit:bug-report` invocation and artifact output
-- Command grid: card-based layout grouped by workflow phase, hover reveals description
-- Smooth scroll, anchor navigation at top
-
-### Page Sections
-
-```
-[Hero — headline + install one-liner + terminal animation]
-[Social proof strip — 2–3 quotes]
-[How It Works — 3-step visual flow]
-[Commands — card grid, grouped by phase]
-[Agents — 3 cards with persona descriptions]
-[v2.0 callout strip — "What's new"]
-[Install — code block + link to GitHub]
-[Footer — license, GitHub link]
-```
-
-### File Layout
-
-```
-docs/
-├── index.html      # Single page, all content inline
-├── style.css       # Extracted styles
-└── assets/
-    └── terminal.svg  # Optional animated terminal demo
-```
-
-### GitHub Pages Activation
-
-Enable via: repo Settings → Pages → Source: `docs/` folder on `main` branch.
-No `_config.yml` needed for plain HTML.
-
----
-
-## Implementation Phases
-
-1. **CHANGELOG.md** — add v2.0 entry (lowest risk, no structural changes)
-2. **CONTRIBUTING.md** — rewrite with v2 conventions
-3. **README.md** — full rewrite with new narrative and all 15 skills
-4. **docs/index.html + style.css** — build GitHub Pages site
-5. **GitHub Pages** — enable in repo settings, verify deploy
-
-## Success Criteria
-
-- All 15 skills documented in README
-- CHANGELOG has a v2.0 entry
-- CONTRIBUTING reflects v2 conventions
-- GitHub Pages site live and visually compelling
-- No broken links or missing content
diff --git a/docs/plans/2026-03-06-docs-github-pages-plan.md b/docs/plans/2026-03-06-docs-github-pages-plan.md
deleted file mode 100644
index 6eea664..0000000
--- a/docs/plans/2026-03-06-docs-github-pages-plan.md
+++ /dev/null
@@ -1,916 +0,0 @@
-# Docs Rewrite + GitHub Pages Implementation Plan
-
-> **For Claude:** REQUIRED SUB-SKILL: Use superpowers:executing-plans to implement this plan task-by-task.
-
-**Goal:** Rewrite README, CHANGELOG, and CONTRIBUTING for v2.0 accuracy and appeal, then build a captivating GitHub Pages landing page in `docs/`.
-
-**Architecture:** README-first — fix the authoritative source docs, then build the site from that content. Five sequential tasks, each committed independently. No build step; GitHub Pages serves `docs/index.html` directly from `main`.
-
-**Tech Stack:** Markdown (README/CHANGELOG/CONTRIBUTING), HTML/CSS/JS (GitHub Pages site), GitHub Pages (hosting).
-
-**Design doc:** `docs/plans/2026-03-06-docs-github-pages-design.md`
-
----
-
-### Task 1: CHANGELOG.md — Add v2.0 Entry
-
-**Files:**
-- Modify: `CHANGELOG.md`
-
-**Context:** The CHANGELOG currently ends at `[1.1.0]`. The v2.0 commit `afaaf2d` is undocumented. v2.0 added: 5 new skills (test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis), redesigned 5 existing skills (e2e-test, bug-report, api-test, regression-planner, pr-review), redesigned 3 agents with tool restrictions/return contracts/persistent memory, a state-aware system, shared references (`skills/shared/references/`), and multi-format output.
-
-**Step 1: Read the current CHANGELOG**
-
-Read `CHANGELOG.md` fully before editing.
-
-**Step 2: Prepend the v2.0 entry**
-
-Add above `## [1.1.0]`:
-
-```markdown
-## [2.0.0] — 2026-03-06
-
-QA Toolkit v2.0 is a ground-up rethink of how the plugin operates. The core idea: every skill now reads your project state before producing output, so artifacts build on each other rather than starting cold every time.
-
-### Added
-- **State-aware skill system** — all skills read `.qa-config.json` on invocation, adapt output to detected stack, and write findings back so subsequent skills have context
-- **Shared references** (`skills/shared/references/`) — `context-preamble.md`, `state-integration.md`, `output-formats.md`, `artifact-organization.md` loaded by all skills to enforce consistent behavior
-- **5 new skills**: `test-plan`, `exploratory-testing`, `coverage-gap`, `risk-prioritization`, `flaky-test-diagnosis`
-- **Multi-format output** — skills produce output in the format best suited to the artifact type (Markdown reports, Gherkin, cURL/Postman/Playwright, JSON/CSV/SQL, etc.)
-- **Redesigned agents** — `qa-reviewer`, `qa-lead`, `qa-explorer` rebuilt with explicit tool restrictions, typed return contracts, and persistent memory across turns
-
-### Changed
-- **5 redesigned skills**: `e2e-test`, `bug-report`, `api-test`, `regression-planner`, `pr-review` — all updated for state-awareness and multi-format output
-- Agent personas now define tool access boundaries (not just tone), making subagent behavior predictable and composable
-
-### Fixed
-- Updated test suite to match v2 skill behavior (stale v1.x assertions removed)
-
-```
-
-**Step 3: Verify**
-
-Read `CHANGELOG.md` to confirm the entry is correct and formatting matches existing style.
-
-**Step 4: Commit**
-
-```bash
-git add CHANGELOG.md
-git commit -s -m "docs: add v2.0 changelog entry"
-```
-
----
-
-### Task 2: CONTRIBUTING.md — Rewrite for v2 Conventions
-
-**Files:**
-- Modify: `CONTRIBUTING.md`
-
-**Context:** Current CONTRIBUTING.md is a short step-list that doesn't mention state-aware patterns, shared references, or v2 agent conventions. Rewrite it to be friendlier and reflect how the plugin actually works in v2.
-
-**Step 1: Read the current CONTRIBUTING.md**
-
-Read `CONTRIBUTING.md` fully.
-
-**Step 2: Also read the shared reference files for accuracy**
-
-Read these to understand what to document:
-- `skills/shared/references/state-integration.md`
-- `skills/shared/references/context-preamble.md`
-- `skills/shared/references/output-formats.md`
-
-**Step 3: Write the new CONTRIBUTING.md**
-
-Full replacement. Content:
-
-```markdown
-# Contributing
-
-## How It All Fits Together
-
-QA Toolkit is a Claude Code plugin. It has three moving parts:
-
-- **Skills** (`skills/<name>/SKILL.md`) — one-shot slash commands that produce structured QA artifacts
-- **Agents** (`agents/<name>.md`) — conversational personas for multi-turn QA work
-- **Scripts** (`scripts/`) — Node.js hooks that run on session start/end (auto-detection, activity logging)
-
-All skills follow the **state-aware pattern**: read project context from `.qa-config.json` → adapt output to the detected stack → write findings back so the next skill has more to work with.
-
-## How to Contribute
-
-1. Fork the repository
-2. Create a feature branch: `git checkout -b feature/your-feature`
-3. Make your changes following the conventions below
-4. Run the test suite: `node --test tests/*.test.js`
-5. Test locally: `claude --plugin-dir ./qa-toolkit`
-6. Submit a pull request
-
-## Adding a New Skill
-
-### Minimum structure
-
-```
-skills/<skill-name>/
-├── SKILL.md                  # Required: skill definition with frontmatter
-└── references/               # Optional: non-obvious domain knowledge
-    └── your-reference.md
-```
-
-### SKILL.md conventions
-
-- YAML frontmatter must have `name` and `description` (one sentence max)
-- Begin with the context preamble: load `skills/shared/references/context-preamble.md` to read project state
-- Adapt output based on detected `languages`, `frameworks`, `testFrameworks` from `.qa-config.json`
-- Save artifacts to `<outputDir>/<category>/` (read `outputDir` from `.qa-config.json`)
-- End with a "Suggested Next Steps" section with conditional cross-references to related skills
-
-### State-aware pattern
-
-Every skill should follow this flow:
-
-1. **Read state** — check if `.qa-config.json` exists; if yes, load it for stack context
-2. **Adapt output** — use detected frameworks to shape format (e.g., Jest vs Pytest syntax, Jira vs GitHub Issues format)
-3. **Write findings** — append relevant findings to a skill-specific log in `<outputDir>/` so subsequent skills have more context
-
-See `skills/shared/references/state-integration.md` for the full pattern with examples.
-
-### Reference files
-
-Put reference files in `skills/<name>/references/` only when they contain **non-obvious, specific guidance** that Claude wouldn't already know. Generic QA knowledge doesn't belong here.
-
-Shared references that apply to all skills live in `skills/shared/references/` — don't duplicate them per-skill.
-
-## Adding a New Agent
-
-```
-agents/<agent-name>.md        # Required: agent definition with frontmatter
-```
-
-### Agent conventions (v2)
-
-- YAML frontmatter must have `name` and `description`
-- Define the **persona and tool access boundaries**, not the workflow — workflows belong in skills
-- Specify **tool restrictions**: which tools the agent may and may not use
-- Define a **return contract**: what the agent always returns at the end of each turn (e.g., a summary, a next-step prompt, a structured finding)
-- Agents have **persistent memory** across turns — design the persona to accumulate context, not start fresh each turn
-
-## Scripts
-
-- All scripts are Node.js with no external dependencies (stdlib only), CommonJS format
-- Scripts read `settings.json` for plugin-level config (`outputDir`, `hooksEnabled`)
-- Hook scripts run on `SessionStart` and `Stop` events
-
-## Testing
-
-```bash
-node --test tests/*.test.js
-```
-
-Tests use Node.js built-in test runner (zero dependencies). When adding a skill or modifying a script, add or update the corresponding test in `tests/`.
-
-## Local Development
-
-```bash
-claude --debug --plugin-dir ./qa-toolkit
-```
-
-The `--debug` flag shows hook execution and skill loading in the Claude Code output.
-```
-
-**Step 4: Verify**
-
-Read `CONTRIBUTING.md` to confirm it's complete and accurate.
-
-**Step 5: Commit**
-
-```bash
-git add CONTRIBUTING.md
-git commit -s -m "docs: rewrite CONTRIBUTING for v2 conventions and friendlier tone"
-```
-
----
-
-### Task 3: README.md — Full Rewrite
-
-**Files:**
-- Modify: `README.md`
-
-**Context:** Full narrative rewrite. Speaks to both developers and QA leads. GSD-inspired: punchy, personal, anti-enterprise. All 15 skills documented, grouped by workflow phase. Badges, social proof, personal founder voice.
-
-**Step 1: Read the current README.md**
-
-Read `README.md` fully before rewriting.
-
-**Step 2: Also read all 15 SKILL.md descriptions for accuracy**
-
-Read the `description` field from frontmatter of each SKILL.md to get accurate one-liners:
-- `skills/setup/SKILL.md`
-- `skills/pr-review/SKILL.md`
-- `skills/bug-report/SKILL.md`
-- `skills/test-cases/SKILL.md`
-- `skills/api-test/SKILL.md`
-- `skills/e2e-test/SKILL.md`
-- `skills/regression-planner/SKILL.md`
-- `skills/accessibility/SKILL.md`
-- `skills/release-readiness/SKILL.md`
-- `skills/test-data/SKILL.md`
-- `skills/test-plan/SKILL.md`
-- `skills/exploratory-testing/SKILL.md`
-- `skills/coverage-gap/SKILL.md`
-- `skills/risk-prioritization/SKILL.md`
-- `skills/flaky-test-diagnosis/SKILL.md`
-
-And agent descriptions:
-- `agents/qa-reviewer.md`
-- `agents/qa-explorer.md`
-- `agents/qa-lead.md`
-
-**Step 3: Write the new README.md**
-
-Full replacement following this structure:
-
-```markdown
-<div align="center">
-
-# QA Toolkit
-
-**QA that thinks for itself.**
-
-Describe what you're testing. Get a structured, stack-aware artifact — instantly.
-
-[![Plugin](https://img.shields.io/badge/claude--plugin-qa--toolkit-blue?style=for-the-badge)](https://github.com/cyberwalk3r/qa-toolkit)
-[![License](https://img.shields.io/badge/license-MIT-green?style=for-the-badge)](LICENSE)
-[![CI](https://img.shields.io/github/actions/workflow/status/cyberwalk3r/qa-toolkit/test.yml?branch=main&style=for-the-badge&label=Tests)](https://github.com/cyberwalk3r/qa-toolkit/actions)
-
-<br>
-
-\`\`\`bash
-claude plugin add github:cyberwalk3r/qa-toolkit
-\`\`\`
-
-<br>
-
-*"Finally — bug reports that don't need three rounds of editing."*
-
-*"I stopped dreading PR reviews. It just does them."*
-
-*"Replaced half my QA templates in a week."*
-
-</div>
-
----
-
-## Why I Built This
-
-QA work is repetitive in the worst way. Every bug report needs the same structure. Every PR review needs the same checks. Every regression plan needs the same risk analysis. And every team member does it differently.
-
-I wanted a tool that handles the formatting so you can focus on the judgment. Describe what you're testing in plain English — get back something you can actually use.
-
-So I built QA Toolkit. It auto-detects your stack, remembers your project context, and produces consistent QA artifacts every time. Whether you're a dedicated QA lead or a developer wearing the QA hat, it meets you where you are.
-
----
-
-## How It Works
-
-**1. Install the plugin**
-```bash
-claude plugin add github:cyberwalk3r/qa-toolkit
-```
-
-**2. Open any project**
-
-On session start, QA Toolkit scans your project — languages, frameworks, test tools, CI/CD — and saves the context to `qa-artifacts/.qa-config.json`. No prompts. No setup.
-
-**3. Describe what you need**
-```
-/qa-toolkit:bug-report The checkout page crashes when I enter a long address
-```
-
-You get back a fully structured bug report: severity classification, reproduction steps, expected vs actual, environment fields, and duplicate search terms — formatted for your bug tracker.
-
----
-
-## Commands
-
-### Authoring
-| Command | What You Get |
-|---------|-------------|
-| `/qa-toolkit:test-cases` | Test cases from requirements — table, Gherkin, or checklist |
-| `/qa-toolkit:test-plan` | Full test plan with scope, strategy, and coverage targets |
-| `/qa-toolkit:test-data` | Synthetic test data — JSON, CSV, or SQL |
-| `/qa-toolkit:e2e-test` | Playwright test scaffold with line-by-line comments |
-| `/qa-toolkit:api-test` | API test suite — cURL, Postman collection, or Playwright |
-
-### Review
-| Command | What You Get |
-|---------|-------------|
-| `/qa-toolkit:pr-review` | Risk-flagged PR review with plain-English summary and QA checklist |
-| `/qa-toolkit:bug-report` | Structured bug report from a casual description |
-| `/qa-toolkit:exploratory-testing` | Exploratory test charters with session-based heuristics |
-
-### Release
-| Command | What You Get |
-|---------|-------------|
-| `/qa-toolkit:regression-planner` | Risk-based regression plan with time estimates |
-| `/qa-toolkit:release-readiness` | Go/no-go assessment with quality gate scoring |
-| `/qa-toolkit:risk-prioritization` | Ranked risk matrix across features and change areas |
-
-### Analysis
-| Command | What You Get |
-|---------|-------------|
-| `/qa-toolkit:coverage-gap` | Coverage gap analysis against requirements or test plan |
-| `/qa-toolkit:flaky-test-diagnosis` | Root cause analysis for flaky tests with fix recommendations |
-| `/qa-toolkit:accessibility` | WCAG 2.1 audit with plain-English manual test scripts |
-| `/qa-toolkit:setup` | Read project docs, confirm detection, save preferences |
-
----
-
-## Agents
-
-Three QA personas for multi-turn interactive work — use these when you want a conversation, not a document.
-
-| Agent | Use When |
-|-------|----------|
-| `qa-reviewer` | Walk me through this PR. What should I test? |
-| `qa-explorer` | Help me find edge cases in this new feature |
-| `qa-lead` | Let's decide if we're ready to release |
-
-Agents remember context across the conversation and adapt as you share more.
-
-### Skills vs Agents
-
-**Skills** (`/qa-toolkit:*`) produce **one-shot structured artifacts** — saved to disk, formatted, done.
-
-**Agents** are **conversational** — use them for ongoing discussion, decision-making, and exploratory work.
-
----
-
-## What's New in v2.0
-
-**State-aware skills.** Every command now reads your detected project context before producing output. It knows you're using Jest, not Pytest. It knows your CI is GitHub Actions. It knows what artifacts you've already generated this session.
-
-**15 skills** (up from 10). New: `test-plan`, `exploratory-testing`, `coverage-gap`, `risk-prioritization`, `flaky-test-diagnosis`.
-
-**Redesigned agents.** `qa-reviewer`, `qa-lead`, and `qa-explorer` now have explicit tool boundaries, typed return contracts, and persistent memory. Their behavior is predictable and composable.
-
----
-
-## Output
-
-Everything saves to `qa-artifacts/` in your project root (configurable via `settings.json`):
-
-```
-qa-artifacts/
-├── .qa-config.json          # Auto-detected project config
-├── .qa-activity.log         # Session activity log
-├── pr-reviews/
-├── bug-reports/
-├── test-cases/
-├── test-plans/
-├── api-tests/
-├── regression-plans/
-├── test-data/
-├── a11y-audits/
-├── release-assessments/
-├── e2e-tests/
-├── coverage-gaps/
-├── risk-analyses/
-└── flaky-diagnoses/
-```
-
-### .qa-config.json Schema
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `detectedAt` | string | ISO 8601 timestamp |
-| `projectRoot` | string | Absolute path to project directory |
-| `outputDir` | string | Configured output directory name |
-| `languages` | string[] | Detected languages |
-| `frameworks` | string[] | Detected frameworks |
-| `testFrameworks` | string[] | Detected test tools |
-| `cicd` | string[] | Detected CI/CD systems |
-| `packageManager` | string\|null | Detected package manager |
-| `hasClaudeMd` | boolean | Whether CLAUDE.md exists |
-| `hasReadme` | boolean | Whether README.md exists |
-| `existingDocs` | object[] | Detected documentation files |
-| `existingTestDirs` | string[] | Detected test directories |
-
-**Note:** `projectRoot` is machine-specific — don't commit `.qa-config.json` to version control. The config is cached for 24 hours; delete it to force re-detection.
-
----
-
-## Permissions & Side Effects
-
-The plugin ships a `.claude/settings.local.json` that pre-approves `Bash(git:*)` for git operations used by the PR review skill. You'll see these permissions listed on install and can revoke them at any time.
-
-**Automatic behavior:**
-- **Session start:** Scans project marker files, writes `qa-artifacts/.qa-config.json`. No network calls.
-- **Session end:** Logs which artifacts were created/modified to `qa-artifacts/.qa-activity.log`. No network calls.
-
-**To disable hooks** without removing the plugin:
-```json
-{
-    "agent": "qa-reviewer",
-    "outputDir": "qa-artifacts",
-    "hooksEnabled": false
-}
-```
-
----
-
-## Install
-
-```bash
-# From GitHub
-claude plugin add github:cyberwalk3r/qa-toolkit
-
-# Local development
-claude plugin add ./qa-toolkit
-```
-
----
-
-## Works With Your Existing Setup
-
-- **Respects existing CLAUDE.md** — reads it for context, never overwrites
-- **Reads your docs** — TESTING.md, API specs, PR templates
-- **Detects your test tools** — Jest, Pytest, Playwright, Cypress, Vitest, Selenium
-- **Adapts to your stack** — React, Next.js, Django, FastAPI, .NET, Spring, Go, and more
-
----
-
-## License
-
-MIT
-```
-
-**Step 4: Verify**
-
-Read `README.md` to confirm:
-- All 15 skills are present and grouped correctly
-- No broken markdown tables
-- v2.0 section is accurate
-- Install commands are correct
-
-**Step 5: Commit**
-
-```bash
-git add README.md
-git commit -s -m "docs: rewrite README for v2.0 with new narrative and all 15 skills"
-```
-
----
-
-### Task 4: GitHub Pages Site
-
-**Files:**
-- Create: `docs/index.html`
-- Create: `docs/style.css`
-
-**Context:** Single-page marketing + docs site. Dark theme, amber accent, terminal aesthetic. Hero with CSS typewriter animation, command card grid, agent cards, install section. No build step — GitHub Pages serves these directly.
-
-**Step 1: Create `docs/style.css`**
-
-```css
-/* Reset & base */
-*, *::before, *::after { box-sizing: border-box; margin: 0; padding: 0; }
-
-:root {
-  --bg: #0d0d0d;
-  --surface: #161616;
-  --border: #2a2a2a;
-  --amber: #f59e0b;
-  --amber-dim: #b45309;
-  --text: #e5e7eb;
-  --text-muted: #6b7280;
-  --code-bg: #1a1a1a;
-  --font-mono: 'JetBrains Mono', 'Fira Code', 'Cascadia Code', monospace;
-  --font-sans: -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;
-}
-
-html { scroll-behavior: smooth; }
-body { background: var(--bg); color: var(--text); font-family: var(--font-sans); line-height: 1.6; }
-
-/* Nav */
-nav {
-  position: fixed; top: 0; left: 0; right: 0; z-index: 100;
-  background: rgba(13,13,13,0.9); backdrop-filter: blur(8px);
-  border-bottom: 1px solid var(--border);
-  padding: 0.75rem 2rem;
-  display: flex; align-items: center; justify-content: space-between;
-}
-nav .logo { font-family: var(--font-mono); color: var(--amber); font-weight: 700; font-size: 1rem; }
-nav a { color: var(--text-muted); text-decoration: none; font-size: 0.875rem; margin-left: 2rem; transition: color 0.2s; }
-nav a:hover { color: var(--amber); }
-
-/* Hero */
-.hero {
-  min-height: 100vh; display: flex; flex-direction: column;
-  align-items: center; justify-content: center;
-  padding: 6rem 2rem 4rem; text-align: center;
-}
-.hero h1 { font-family: var(--font-mono); font-size: clamp(2rem, 5vw, 3.5rem); font-weight: 700; letter-spacing: -0.02em; line-height: 1.1; }
-.hero h1 span { color: var(--amber); }
-.hero .tagline { font-size: clamp(1rem, 2vw, 1.25rem); color: var(--text-muted); margin: 1.5rem 0 2.5rem; max-width: 560px; }
-
-/* Install block */
-.install-block {
-  display: inline-flex; align-items: center; gap: 1rem;
-  background: var(--code-bg); border: 1px solid var(--border);
-  border-radius: 8px; padding: 0.875rem 1.5rem;
-  font-family: var(--font-mono); font-size: 0.9rem;
-  margin-bottom: 3rem;
-}
-.install-block .prompt { color: var(--amber); }
-.copy-btn {
-  background: none; border: 1px solid var(--border); border-radius: 4px;
-  color: var(--text-muted); cursor: pointer; padding: 0.25rem 0.6rem;
-  font-size: 0.75rem; transition: all 0.2s;
-}
-.copy-btn:hover { border-color: var(--amber); color: var(--amber); }
-
-/* Terminal demo */
-.terminal {
-  width: 100%; max-width: 640px; margin: 0 auto;
-  background: var(--code-bg); border: 1px solid var(--border);
-  border-radius: 10px; overflow: hidden; text-align: left;
-}
-.terminal-bar {
-  background: var(--surface); padding: 0.75rem 1rem;
-  display: flex; align-items: center; gap: 0.5rem;
-  border-bottom: 1px solid var(--border);
-}
-.dot { width: 12px; height: 12px; border-radius: 50%; }
-.dot.red { background: #ff5f57; }
-.dot.yellow { background: #febc2e; }
-.dot.green { background: #28c840; }
-.terminal-title { font-size: 0.75rem; color: var(--text-muted); margin-left: auto; font-family: var(--font-mono); }
-.terminal-body { padding: 1.25rem 1.5rem; font-family: var(--font-mono); font-size: 0.825rem; line-height: 1.7; }
-.line-cmd { color: var(--amber); }
-.line-out { color: #6b7280; }
-.line-key { color: #34d399; }
-.line-val { color: var(--text); }
-.cursor { display: inline-block; width: 2px; height: 1em; background: var(--amber); animation: blink 1s step-end infinite; vertical-align: text-bottom; }
-@keyframes blink { 50% { opacity: 0; } }
-
-/* Typewriter */
-.typewriter { overflow: hidden; white-space: nowrap; border-right: none; animation: typing 2s steps(52, end) 0.5s both; }
-@keyframes typing { from { width: 0; } to { width: 100%; } }
-
-/* Quotes */
-.quotes { background: var(--surface); border-top: 1px solid var(--border); border-bottom: 1px solid var(--border); padding: 3rem 2rem; }
-.quotes-inner { max-width: 900px; margin: 0 auto; display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); gap: 2rem; }
-.quote { padding: 1.5rem; border: 1px solid var(--border); border-radius: 8px; }
-.quote p { color: var(--text); font-size: 0.9rem; line-height: 1.6; margin-bottom: 1rem; }
-.quote cite { font-size: 0.775rem; color: var(--text-muted); font-style: normal; }
-
-/* Sections */
-section { padding: 5rem 2rem; }
-.container { max-width: 960px; margin: 0 auto; }
-.section-label { font-family: var(--font-mono); font-size: 0.75rem; color: var(--amber); text-transform: uppercase; letter-spacing: 0.1em; margin-bottom: 0.75rem; }
-h2 { font-size: clamp(1.5rem, 3vw, 2rem); font-weight: 700; margin-bottom: 1rem; }
-.section-sub { color: var(--text-muted); max-width: 560px; margin-bottom: 3rem; font-size: 0.95rem; }
-
-/* How it works */
-.steps { display: grid; grid-template-columns: repeat(auto-fit, minmax(240px, 1fr)); gap: 1.5rem; }
-.step { padding: 1.75rem; border: 1px solid var(--border); border-radius: 8px; position: relative; }
-.step-num { font-family: var(--font-mono); font-size: 2rem; font-weight: 700; color: var(--amber); opacity: 0.3; margin-bottom: 0.75rem; }
-.step h3 { font-size: 1rem; margin-bottom: 0.5rem; }
-.step p { font-size: 0.875rem; color: var(--text-muted); }
-
-/* Commands */
-.phase-group { margin-bottom: 2.5rem; }
-.phase-label { font-family: var(--font-mono); font-size: 0.7rem; color: var(--amber); text-transform: uppercase; letter-spacing: 0.12em; margin-bottom: 1rem; padding-bottom: 0.5rem; border-bottom: 1px solid var(--border); }
-.command-grid { display: grid; grid-template-columns: repeat(auto-fill, minmax(280px, 1fr)); gap: 1rem; }
-.command-card {
-  padding: 1.25rem; border: 1px solid var(--border); border-radius: 8px;
-  transition: border-color 0.2s, background 0.2s; cursor: default;
-}
-.command-card:hover { border-color: var(--amber-dim); background: var(--surface); }
-.command-card code { font-family: var(--font-mono); font-size: 0.775rem; color: var(--amber); display: block; margin-bottom: 0.5rem; }
-.command-card p { font-size: 0.825rem; color: var(--text-muted); }
-
-/* Agents */
-.agent-grid { display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); gap: 1.5rem; }
-.agent-card { padding: 1.75rem; border: 1px solid var(--border); border-radius: 8px; }
-.agent-card h3 { font-family: var(--font-mono); color: var(--amber); margin-bottom: 0.5rem; font-size: 0.9rem; }
-.agent-card .role { font-size: 0.875rem; font-weight: 600; margin-bottom: 0.5rem; }
-.agent-card p { font-size: 0.825rem; color: var(--text-muted); }
-
-/* v2 callout */
-.v2-callout {
-  background: linear-gradient(135deg, rgba(245,158,11,0.08), rgba(245,158,11,0.02));
-  border: 1px solid rgba(245,158,11,0.3); border-radius: 12px;
-  padding: 2.5rem; margin-top: 2rem;
-}
-.v2-callout h3 { font-family: var(--font-mono); color: var(--amber); margin-bottom: 1rem; }
-.v2-callout ul { list-style: none; display: grid; grid-template-columns: repeat(auto-fit, minmax(200px, 1fr)); gap: 0.5rem; }
-.v2-callout li { font-size: 0.875rem; color: var(--text-muted); padding-left: 1rem; position: relative; }
-.v2-callout li::before { content: '→'; position: absolute; left: 0; color: var(--amber); }
-
-/* Install section */
-.install-section { background: var(--surface); border-top: 1px solid var(--border); }
-.code-block {
-  background: var(--code-bg); border: 1px solid var(--border); border-radius: 8px;
-  padding: 1.25rem 1.5rem; font-family: var(--font-mono); font-size: 0.85rem;
-  color: var(--text); margin-bottom: 1rem; white-space: pre;
-}
-.code-block .comment { color: var(--text-muted); }
-.code-block .cmd { color: var(--amber); }
-
-/* Footer */
-footer {
-  border-top: 1px solid var(--border); padding: 2rem;
-  text-align: center; color: var(--text-muted); font-size: 0.8rem;
-}
-footer a { color: var(--text-muted); text-decoration: none; }
-footer a:hover { color: var(--amber); }
-
-/* Responsive */
-@media (max-width: 640px) {
-  nav a { display: none; }
-  .install-block { flex-direction: column; gap: 0.5rem; }
-}
-```
-
-**Step 2: Create `docs/index.html`**
-
-```html
-<!DOCTYPE html>
-<html lang="en">
-<head>
-  <meta charset="UTF-8">
-  <meta name="viewport" content="width=device-width, initial-scale=1.0">
-  <title>QA Toolkit — QA that thinks for itself</title>
-  <meta name="description" content="A Claude Code plugin that auto-detects your stack and produces structured QA artifacts from plain English descriptions.">
-  <link rel="stylesheet" href="style.css">
-</head>
-<body>
-
-<nav>
-  <span class="logo">/qa-toolkit</span>
-  <div>
-    <a href="#how-it-works">How It Works</a>
-    <a href="#commands">Commands</a>
-    <a href="#agents">Agents</a>
-    <a href="#install">Install</a>
-    <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank">GitHub</a>
-  </div>
-</nav>
-
-<!-- Hero -->
-<section class="hero">
-  <h1>QA that <span>thinks for itself.</span></h1>
-  <p class="tagline">Describe what you're testing. Get a structured, stack-aware artifact — instantly. No templates. No formatting. No busywork.</p>
-
-  <div class="install-block">
-    <span class="prompt">$</span>
-    <span id="install-cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
-    <button class="copy-btn" onclick="copyInstall()">copy</button>
-  </div>
-
-  <div class="terminal" aria-label="Terminal demo">
-    <div class="terminal-bar">
-      <span class="dot red"></span>
-      <span class="dot yellow"></span>
-      <span class="dot green"></span>
-      <span class="terminal-title">Claude Code</span>
-    </div>
-    <div class="terminal-body">
-      <div class="line-cmd typewriter">&gt; /qa-toolkit:bug-report The checkout page crashes with long addresses</div>
-      <br>
-      <div class="line-out" style="animation-delay:2.5s">Detecting project context...</div>
-      <div class="line-key">Stack: <span class="line-val">React · Node.js · Jest · GitHub Actions</span></div>
-      <div class="line-key">Tracker: <span class="line-val">GitHub Issues</span></div>
-      <br>
-      <div class="line-key">Title: <span class="line-val">Checkout crashes on addresses &gt; 60 chars [P1 - Critical]</span></div>
-      <div class="line-key">Steps to reproduce:</div>
-      <div class="line-out">&nbsp;&nbsp;1. Navigate to /checkout</div>
-      <div class="line-out">&nbsp;&nbsp;2. Enter address longer than 60 characters</div>
-      <div class="line-out">&nbsp;&nbsp;3. Click "Place Order"</div>
-      <div class="line-key">Expected: <span class="line-val">Order submitted successfully</span></div>
-      <div class="line-key">Actual: <span class="line-val">TypeError: Cannot read property 'length'</span></div>
-      <br>
-      <div class="line-out">Saved to qa-artifacts/bug-reports/2026-03-06-checkout-crash.md</div>
-      <span class="cursor"></span>
-    </div>
-  </div>
-</section>
-
-<!-- Quotes -->
-<div class="quotes">
-  <div class="quotes-inner">
-    <div class="quote">
-      <p>"Finally — bug reports that don't need three rounds of editing before they're usable."</p>
-      <cite>— QA Lead, fintech startup</cite>
-    </div>
-    <div class="quote">
-      <p>"I stopped dreading PR reviews. It reads the diff, flags the risks, writes the checklist. I just review the output."</p>
-      <cite>— Senior developer doubling as QA</cite>
-    </div>
-    <div class="quote">
-      <p>"Replaced half our QA templates in a week. The team is actually using them consistently for the first time."</p>
-      <cite>— Engineering manager</cite>
-    </div>
-  </div>
-</div>
-
-<!-- How It Works -->
-<section id="how-it-works">
-  <div class="container">
-    <div class="section-label">How It Works</div>
-    <h2>Three steps. That's it.</h2>
-    <p class="section-sub">Install once. Open any project. It auto-detects your stack and gets to work.</p>
-    <div class="steps">
-      <div class="step">
-        <div class="step-num">01</div>
-        <h3>Install the plugin</h3>
-        <p>One command. Works with any project — React, Django, .NET, Go, whatever you're building.</p>
-      </div>
-      <div class="step">
-        <div class="step-num">02</div>
-        <h3>Open your project</h3>
-        <p>On session start, QA Toolkit scans your stack and saves context. Knows your frameworks, test tools, and CI setup automatically.</p>
-      </div>
-      <div class="step">
-        <div class="step-num">03</div>
-        <h3>Describe what you need</h3>
-        <p>Type a slash command in plain English. Get back a structured, formatted artifact saved to disk.</p>
-      </div>
-    </div>
-  </div>
-</section>
-
-<!-- Commands -->
-<section id="commands" style="background: var(--surface); border-top: 1px solid var(--border); border-bottom: 1px solid var(--border);">
-  <div class="container">
-    <div class="section-label">Commands</div>
-    <h2>15 slash commands. Every QA workflow covered.</h2>
-    <p class="section-sub">Grouped by where they fit in your workflow. Each one produces a formatted artifact saved to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/</code>.</p>
-
-    <div class="phase-group">
-      <div class="phase-label">Authoring</div>
-      <div class="command-grid">
-        <div class="command-card"><code>/qa-toolkit:test-cases</code><p>Test cases from requirements — table, Gherkin, or checklist format</p></div>
-        <div class="command-card"><code>/qa-toolkit:test-plan</code><p>Full test plan with scope, strategy, and coverage targets</p></div>
-        <div class="command-card"><code>/qa-toolkit:test-data</code><p>Synthetic test data — JSON, CSV, or SQL</p></div>
-        <div class="command-card"><code>/qa-toolkit:e2e-test</code><p>Playwright test scaffold with line-by-line comments</p></div>
-        <div class="command-card"><code>/qa-toolkit:api-test</code><p>API test suite — cURL, Postman collection, or Playwright</p></div>
-      </div>
-    </div>
-
-    <div class="phase-group">
-      <div class="phase-label">Review</div>
-      <div class="command-grid">
-        <div class="command-card"><code>/qa-toolkit:pr-review</code><p>Risk-flagged PR review with plain-English summary and QA checklist</p></div>
-        <div class="command-card"><code>/qa-toolkit:bug-report</code><p>Structured bug report from a casual description</p></div>
-        <div class="command-card"><code>/qa-toolkit:exploratory-testing</code><p>Exploratory test charters with session-based heuristics</p></div>
-      </div>
-    </div>
-
-    <div class="phase-group">
-      <div class="phase-label">Release</div>
-      <div class="command-grid">
-        <div class="command-card"><code>/qa-toolkit:regression-planner</code><p>Risk-based regression plan with time estimates</p></div>
-        <div class="command-card"><code>/qa-toolkit:release-readiness</code><p>Go/no-go assessment with quality gate scoring</p></div>
-        <div class="command-card"><code>/qa-toolkit:risk-prioritization</code><p>Ranked risk matrix across features and change areas</p></div>
-      </div>
-    </div>
-
-    <div class="phase-group">
-      <div class="phase-label">Analysis</div>
-      <div class="command-grid">
-        <div class="command-card"><code>/qa-toolkit:coverage-gap</code><p>Coverage gap analysis against requirements or test plan</p></div>
-        <div class="command-card"><code>/qa-toolkit:flaky-test-diagnosis</code><p>Root cause analysis for flaky tests with fix recommendations</p></div>
-        <div class="command-card"><code>/qa-toolkit:accessibility</code><p>WCAG 2.1 audit with plain-English manual test scripts</p></div>
-        <div class="command-card"><code>/qa-toolkit:setup</code><p>Read project docs, confirm detection, save preferences</p></div>
-      </div>
-    </div>
-  </div>
-</section>
-
-<!-- Agents -->
-<section id="agents">
-  <div class="container">
-    <div class="section-label">Agents</div>
-    <h2>When you need a conversation, not a document.</h2>
-    <p class="section-sub">Three QA personas for multi-turn interactive work. They remember context across the conversation and adapt as you share more.</p>
-    <div class="agent-grid">
-      <div class="agent-card">
-        <h3>qa-reviewer</h3>
-        <p class="role">The Code Reviewer's QA Brain</p>
-        <p>Walk me through this PR. What should I actually be testing? Translates code changes into testing impact in plain English.</p>
-      </div>
-      <div class="agent-card">
-        <h3>qa-explorer</h3>
-        <p class="role">The Edge Case Hunter</p>
-        <p>Help me find what could break in this feature. Generates exploratory test charters and surfaces edge cases you wouldn't think to test.</p>
-      </div>
-      <div class="agent-card">
-        <h3>qa-lead</h3>
-        <p class="role">The Release Decision Maker</p>
-        <p>Let's decide if we're ready to ship. Makes release decisions, plans regression scope, and produces executive summaries.</p>
-      </div>
-    </div>
-
-    <div class="v2-callout">
-      <h3>What's New in v2.0</h3>
-      <ul>
-        <li>State-aware skills — every command reads your detected project context</li>
-        <li>15 skills — 5 new in v2: test-plan, exploratory-testing, coverage-gap, risk-prioritization, flaky-test-diagnosis</li>
-        <li>Redesigned agents — explicit tool boundaries, typed return contracts, persistent memory</li>
-        <li>Multi-format output — artifacts in the format that actually fits the deliverable</li>
-        <li>Shared references — consistent behavior across all skills</li>
-      </ul>
-    </div>
-  </div>
-</section>
-
-<!-- Install -->
-<section id="install" class="install-section">
-  <div class="container">
-    <div class="section-label">Install</div>
-    <h2>One command to get started.</h2>
-    <p class="section-sub">Works with any project. No configuration required.</p>
-    <div class="code-block"><span class="comment"># From GitHub</span>
-<span class="cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
-
-<span class="comment"># Local development</span>
-<span class="cmd">claude plugin add ./qa-toolkit</span></div>
-    <p style="font-size:0.875rem;color:var(--text-muted)">On first session, auto-detection runs and saves your project context to <code style="color:var(--amber);font-family:var(--font-mono);font-size:0.85em">qa-artifacts/.qa-config.json</code>. Everything is ready.</p>
-  </div>
-</section>
-
-<footer>
-  <p>MIT License &nbsp;·&nbsp; <a href="https://github.com/cyberwalk3r/qa-toolkit" target="_blank">GitHub</a> &nbsp;·&nbsp; Built for Claude Code</p>
-</footer>
-
-<script>
-function copyInstall() {
-  const text = document.getElementById('install-cmd').textContent;
-  navigator.clipboard.writeText(text).then(() => {
-    const btn = document.querySelector('.copy-btn');
-    btn.textContent = 'copied!';
-    setTimeout(() => btn.textContent = 'copy', 2000);
-  });
-}
-</script>
-
-</body>
-</html>
-```
-
-**Step 3: Verify locally**
-
-Open `docs/index.html` in a browser (or use `python3 -m http.server 8080 --directory docs`) and check:
-- Hero renders correctly with typewriter animation
-- All 15 command cards are present and grouped
-- 3 agent cards show
-- Install section is correct
-- No broken layout on mobile width
-
-**Step 4: Commit**
-
-```bash
-git add docs/index.html docs/style.css
-git commit -s -m "feat: add GitHub Pages site with landing page and full command reference"
-```
-
----
-
-### Task 5: Enable GitHub Pages
-
-**Files:** None (GitHub repo settings)
-
-**Step 1: Push to remote**
-
-```bash
-git push origin main
-```
-
-**Step 2: Enable GitHub Pages**
-
-In the GitHub repo:
-1. Go to Settings → Pages
-2. Source: Deploy from branch
-3. Branch: `main`, folder: `/docs`
-4. Click Save
-
-**Step 3: Verify**
-
-After ~60 seconds, the site will be live at `https://cyberwalk3r.github.io/qa-toolkit/`.
-
-Check:
-- Site loads
-- Navigation links work
-- Terminal animation plays
-- Install copy button works
-
-**Step 4: Add site URL to README**
-
-Add the Pages URL to `README.md` — either as a badge or a link in the header section.
-
-```bash
-git add README.md
-git commit -s -m "docs: add GitHub Pages site link to README"
-```
diff --git a/docs/style.css b/docs/style.css
index 46422ed..57c6685 100644
--- a/docs/style.css
+++ b/docs/style.css
@@ -3,157 +3,597 @@
 
 :root {
   --bg: #0d0d0d;
-  --surface: #161616;
-  --border: #2a2a2a;
+  --surface: #141414;
+  --surface-2: #1a1a1a;
+  --border: #252525;
+  --border-hover: rgba(245,158,11,0.45);
   --amber: #f59e0b;
-  --amber-dim: #b45309;
+  --amber-dim: #d97706;
+  --amber-glow: rgba(245,158,11,0.12);
+  --green: #34d399;
   --text: #e5e7eb;
   --text-muted: #6b7280;
-  --code-bg: #1a1a1a;
+  --code-bg: #111111;
   --font-mono: 'JetBrains Mono', 'Fira Code', 'Cascadia Code', monospace;
   --font-sans: -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;
+  --radius: 10px;
+  --radius-sm: 6px;
+  --ease: cubic-bezier(0.22, 1, 0.36, 1);
 }
 
 html { scroll-behavior: smooth; }
-body { background: var(--bg); color: var(--text); font-family: var(--font-sans); line-height: 1.6; }
+body {
+  background: var(--bg);
+  color: var(--text);
+  font-family: var(--font-sans);
+  line-height: 1.6;
+  overflow-x: hidden;
+}
 
-/* Nav */
+/* ===================== NAV ===================== */
 nav {
   position: fixed; top: 0; left: 0; right: 0; z-index: 100;
-  background: rgba(13,13,13,0.9); backdrop-filter: blur(8px);
+  background: rgba(13,13,13,0.82);
+  backdrop-filter: blur(16px) saturate(180%);
+  -webkit-backdrop-filter: blur(16px) saturate(180%);
   border-bottom: 1px solid var(--border);
   padding: 0.75rem 2rem;
   display: flex; align-items: center; justify-content: space-between;
 }
-nav .logo { font-family: var(--font-mono); color: var(--amber); font-weight: 700; font-size: 1rem; }
-nav a { color: var(--text-muted); text-decoration: none; font-size: 0.875rem; margin-left: 2rem; transition: color 0.2s; }
-nav a:hover { color: var(--amber); }
+nav .logo {
+  font-family: var(--font-mono);
+  color: var(--amber);
+  font-weight: 700;
+  font-size: 1rem;
+  letter-spacing: -0.02em;
+}
+.nav-links a {
+  color: var(--text-muted);
+  text-decoration: none;
+  font-size: 0.875rem;
+  margin-left: 2rem;
+  transition: color 0.18s ease;
+}
+.nav-links a:hover { color: var(--amber); }
 
-/* Hero */
+/* ===================== HERO ===================== */
 .hero {
-  min-height: 100vh; display: flex; flex-direction: column;
-  align-items: center; justify-content: center;
-  padding: 6rem 2rem 4rem; text-align: center;
+  min-height: 100vh;
+  position: relative;
+  overflow: hidden;
+  display: flex;
+  flex-direction: column;
+  align-items: center;
+  justify-content: center;
+  padding: 6rem 2rem 5rem;
+  text-align: center;
+}
+
+/* Dot grid background */
+.hero-grid {
+  position: absolute; inset: 0;
+  background-image: radial-gradient(circle at 1px 1px, rgba(245,158,11,0.13) 1px, transparent 0);
+  background-size: 32px 32px;
+  pointer-events: none;
+  mask-image: radial-gradient(ellipse 85% 85% at 50% 50%, black 30%, transparent 100%);
+  -webkit-mask-image: radial-gradient(ellipse 85% 85% at 50% 50%, black 30%, transparent 100%);
+}
+
+/* Ambient amber glow */
+.hero-glow {
+  position: absolute;
+  width: 900px; height: 900px;
+  background: radial-gradient(circle, rgba(245,158,11,0.09) 0%, transparent 65%);
+  top: 50%; left: 50%;
+  transform: translate(-50%, -50%);
+  pointer-events: none;
+  animation: pulse-glow 7s ease-in-out infinite;
+}
+@keyframes pulse-glow {
+  0%, 100% { opacity: 0.6; transform: translate(-50%, -50%) scale(1); }
+  50% { opacity: 1; transform: translate(-50%, -50%) scale(1.1); }
+}
+
+.hero-content {
+  position: relative;
+  z-index: 1;
+  width: 100%;
+  max-width: 820px;
+}
+
+/* Badge */
+.hero-badge {
+  display: inline-block;
+  font-family: var(--font-mono);
+  font-size: 0.7rem;
+  color: var(--amber);
+  border: 1px solid rgba(245,158,11,0.3);
+  background: rgba(245,158,11,0.06);
+  border-radius: 100px;
+  padding: 0.3rem 1rem;
+  margin-bottom: 1.5rem;
+  letter-spacing: 0.06em;
+  animation: fade-up 0.6s var(--ease) both;
+}
+
+.hero h1 {
+  font-family: var(--font-mono);
+  font-size: clamp(2.1rem, 5.5vw, 4rem);
+  font-weight: 700;
+  letter-spacing: -0.03em;
+  line-height: 1.07;
+  animation: fade-up 0.65s var(--ease) 0.1s both;
+}
+
+/* Shimmer gradient text */
+.shimmer {
+  background: linear-gradient(
+    90deg,
+    #f59e0b 0%,
+    #fde68a 25%,
+    #f59e0b 45%,
+    #d97706 65%,
+    #fbbf24 80%,
+    #f59e0b 100%
+  );
+  background-size: 300% auto;
+  -webkit-background-clip: text;
+  -webkit-text-fill-color: transparent;
+  background-clip: text;
+  animation: shimmer 5s linear infinite;
+}
+@keyframes shimmer {
+  0%   { background-position: 0% center; }
+  100% { background-position: 300% center; }
+}
+
+.hero .tagline {
+  font-size: clamp(0.95rem, 2vw, 1.2rem);
+  color: var(--text-muted);
+  margin: 1.25rem auto 2rem;
+  max-width: 520px;
+  animation: fade-up 0.65s var(--ease) 0.18s both;
 }
-.hero h1 { font-family: var(--font-mono); font-size: clamp(2rem, 5vw, 3.5rem); font-weight: 700; letter-spacing: -0.02em; line-height: 1.1; }
-.hero h1 span { color: var(--amber); }
-.hero .tagline { font-size: clamp(1rem, 2vw, 1.25rem); color: var(--text-muted); margin: 1.5rem 0 2.5rem; max-width: 560px; }
 
 /* Install block */
 .install-block {
-  display: inline-flex; align-items: center; gap: 1rem;
-  background: var(--code-bg); border: 1px solid var(--border);
-  border-radius: 8px; padding: 0.875rem 1.5rem;
-  font-family: var(--font-mono); font-size: 0.9rem;
-  margin-bottom: 3rem;
+  display: inline-flex;
+  align-items: center;
+  gap: 1rem;
+  background: var(--code-bg);
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  padding: 0.875rem 1.5rem;
+  font-family: var(--font-mono);
+  font-size: 0.875rem;
+  margin-bottom: 2rem;
+  animation: fade-up 0.65s var(--ease) 0.26s both;
+  transition: border-color 0.2s ease, box-shadow 0.2s ease;
+}
+.install-block:hover {
+  border-color: var(--border-hover);
+  box-shadow: 0 0 28px rgba(245,158,11,0.1);
 }
 .install-block .prompt { color: var(--amber); }
+
 .copy-btn {
-  background: none; border: 1px solid var(--border); border-radius: 4px;
-  color: var(--text-muted); cursor: pointer; padding: 0.25rem 0.6rem;
-  font-size: 0.75rem; transition: all 0.2s;
+  background: none;
+  border: 1px solid var(--border);
+  border-radius: var(--radius-sm);
+  color: var(--text-muted);
+  cursor: pointer;
+  padding: 0.25rem 0.7rem;
+  font-family: var(--font-mono);
+  font-size: 0.7rem;
+  transition: border-color 0.18s ease, color 0.18s ease;
+  white-space: nowrap;
+}
+.copy-btn:hover,
+.copy-btn.copied { border-color: var(--amber); color: var(--amber); }
+
+/* Stats strip */
+.stats-strip {
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  gap: 1.75rem;
+  margin-bottom: 2.5rem;
+  animation: fade-up 0.65s var(--ease) 0.34s both;
+}
+.stat { text-align: center; }
+.stat-num {
+  display: block;
+  font-family: var(--font-mono);
+  font-size: 1.6rem;
+  font-weight: 700;
+  color: var(--amber);
+  line-height: 1;
+}
+.stat-label {
+  display: block;
+  font-size: 0.68rem;
+  color: var(--text-muted);
+  text-transform: uppercase;
+  letter-spacing: 0.09em;
+  margin-top: 0.25rem;
+}
+.stat-divider {
+  width: 1px;
+  height: 34px;
+  background: var(--border);
+  flex-shrink: 0;
 }
-.copy-btn:hover { border-color: var(--amber); color: var(--amber); }
 
-/* Terminal demo */
+/* ===================== TERMINAL ===================== */
 .terminal {
-  width: 100%; max-width: 640px; margin: 0 auto;
-  background: var(--code-bg); border: 1px solid var(--border);
-  border-radius: 10px; overflow: hidden; text-align: left;
+  width: 100%;
+  max-width: 660px;
+  margin: 0 auto;
+  background: var(--code-bg);
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  overflow: hidden;
+  text-align: left;
+  animation: fade-up 0.7s var(--ease) 0.44s both;
+  box-shadow:
+    0 32px 80px rgba(0,0,0,0.6),
+    0 0 0 1px rgba(245,158,11,0.05),
+    0 0 60px rgba(245,158,11,0.04);
+  transition: opacity 0.35s ease;
 }
 .terminal-bar {
-  background: var(--surface); padding: 0.75rem 1rem;
-  display: flex; align-items: center; gap: 0.5rem;
+  background: var(--surface);
+  padding: 0.75rem 1rem;
+  display: flex;
+  align-items: center;
+  gap: 0.5rem;
   border-bottom: 1px solid var(--border);
 }
-.dot { width: 12px; height: 12px; border-radius: 50%; }
-.dot.red { background: #ff5f57; }
+.dot { width: 12px; height: 12px; border-radius: 50%; flex-shrink: 0; }
+.dot.red    { background: #ff5f57; }
 .dot.yellow { background: #febc2e; }
-.dot.green { background: #28c840; }
-.terminal-title { font-size: 0.75rem; color: var(--text-muted); margin-left: auto; font-family: var(--font-mono); }
-.terminal-body { padding: 1.25rem 1.5rem; font-family: var(--font-mono); font-size: 0.825rem; line-height: 1.7; }
-.line-cmd { color: var(--amber); }
-.line-out { color: #6b7280; }
-.line-key { color: #34d399; }
-.line-val { color: var(--text); }
-.cursor { display: inline-block; width: 2px; height: 1em; background: var(--amber); animation: blink 1s step-end infinite; vertical-align: text-bottom; }
+.dot.green  { background: #28c840; }
+.terminal-title {
+  font-size: 0.72rem;
+  color: var(--text-muted);
+  margin-left: auto;
+  font-family: var(--font-mono);
+}
+.terminal-body {
+  padding: 1.25rem 1.5rem;
+  font-family: var(--font-mono);
+  font-size: 0.82rem;
+  line-height: 1.8;
+  min-height: 190px;
+}
+.line-cmd   { color: var(--amber); }
+.line-out   { color: #6b7280; }
+.line-key   { color: var(--green); }
+.line-val   { color: var(--text); }
+.line-amber { color: var(--amber); }
+.line-spacer { height: 0.6rem; }
+
+.cursor {
+  display: inline-block;
+  width: 2px;
+  height: 1em;
+  background: var(--amber);
+  animation: blink 1s step-end infinite;
+  vertical-align: text-bottom;
+  margin-left: 2px;
+}
 @keyframes blink { 50% { opacity: 0; } }
 
-/* Typewriter */
-.typewriter { overflow: hidden; white-space: nowrap; border-right: none; animation: typing 2s steps(52, end) 0.5s both; }
+.typewriter {
+  overflow: hidden;
+  white-space: nowrap;
+  animation: typing 2.2s steps(62, end) 0.5s both;
+}
 @keyframes typing { from { width: 0; } to { width: 100%; } }
 
-/* Quotes */
-.quotes { background: var(--surface); border-top: 1px solid var(--border); border-bottom: 1px solid var(--border); padding: 3rem 2rem; }
-.quotes-inner { max-width: 900px; margin: 0 auto; display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); gap: 2rem; }
-.quote { padding: 1.5rem; border: 1px solid var(--border); border-radius: 8px; }
-.quote p { color: var(--text-muted); font-size: 0.9rem; line-height: 1.6; }
-.quote p strong { color: var(--text); display: block; margin-bottom: 0.5rem; font-size: 0.95rem; }
-
-/* Sections */
-section { padding: 5rem 2rem; }
-.container { max-width: 960px; margin: 0 auto; }
-.section-label { font-family: var(--font-mono); font-size: 0.75rem; color: var(--amber); text-transform: uppercase; letter-spacing: 0.1em; margin-bottom: 0.75rem; }
-h2 { font-size: clamp(1.5rem, 3vw, 2rem); font-weight: 700; margin-bottom: 1rem; }
-.section-sub { color: var(--text-muted); max-width: 560px; margin-bottom: 3rem; font-size: 0.95rem; }
-
-/* How it works */
-.steps { display: grid; grid-template-columns: repeat(auto-fit, minmax(240px, 1fr)); gap: 1.5rem; }
-.step { padding: 1.75rem; border: 1px solid var(--border); border-radius: 8px; position: relative; }
-.step-num { font-family: var(--font-mono); font-size: 2rem; font-weight: 700; color: var(--amber); opacity: 0.3; margin-bottom: 0.75rem; }
-.step h3 { font-size: 1rem; margin-bottom: 0.5rem; }
-.step p { font-size: 0.875rem; color: var(--text-muted); }
-
-/* Commands */
-.phase-group { margin-bottom: 2.5rem; }
-.phase-label { font-family: var(--font-mono); font-size: 0.7rem; color: var(--amber); text-transform: uppercase; letter-spacing: 0.12em; margin-bottom: 1rem; padding-bottom: 0.5rem; border-bottom: 1px solid var(--border); }
-.command-grid { display: grid; grid-template-columns: repeat(auto-fill, minmax(280px, 1fr)); gap: 1rem; }
+.hero-hint {
+  margin-top: 1.75rem;
+  font-size: 0.78rem;
+  color: var(--text-muted);
+  animation: fade-hint 3s var(--ease) both;
+}
+@keyframes fade-hint {
+  0%, 60% { opacity: 0; transform: translateY(6px); }
+  100%     { opacity: 1; transform: translateY(0); }
+}
+
+/* ===================== SHARED ANIMATIONS ===================== */
+@keyframes fade-in { from { opacity: 0; } to { opacity: 1; } }
+@keyframes fade-up {
+  from { opacity: 0; transform: translateY(22px); }
+  to   { opacity: 1; transform: translateY(0); }
+}
+
+/* Scroll reveal */
+[data-animate] {
+  opacity: 0;
+  transform: translateY(28px);
+  transition: opacity 0.6s var(--ease), transform 0.6s var(--ease);
+}
+[data-animate].visible {
+  opacity: 1;
+  transform: translateY(0);
+}
+
+/* ===================== VALUE PROPS ===================== */
+.quotes {
+  background: var(--surface);
+  border-top: 1px solid var(--border);
+  border-bottom: 1px solid var(--border);
+  padding: 4rem 2rem;
+}
+.quotes-inner {
+  max-width: 980px;
+  margin: 0 auto;
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(260px, 1fr));
+  gap: 1.25rem;
+}
+.quote {
+  padding: 1.75rem;
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  background: var(--bg);
+  transition: border-color 0.2s ease, box-shadow 0.2s ease, transform 0.2s ease;
+}
+.quote:hover {
+  border-color: rgba(245,158,11,0.3);
+  box-shadow: 0 0 28px rgba(245,158,11,0.07);
+  transform: translateY(-3px);
+}
+.quote p { color: var(--text-muted); font-size: 0.875rem; line-height: 1.75; }
+.quote p strong {
+  color: var(--text);
+  display: block;
+  margin-bottom: 0.5rem;
+  font-size: 0.95rem;
+}
+
+/* ===================== SECTIONS ===================== */
+section { padding: 5.5rem 2rem; }
+.container { max-width: 980px; margin: 0 auto; }
+.section-label {
+  font-family: var(--font-mono);
+  font-size: 0.68rem;
+  color: var(--amber);
+  text-transform: uppercase;
+  letter-spacing: 0.14em;
+  margin-bottom: 0.75rem;
+}
+h2 {
+  font-size: clamp(1.5rem, 3vw, 2.15rem);
+  font-weight: 700;
+  margin-bottom: 1rem;
+  letter-spacing: -0.025em;
+}
+.section-sub {
+  color: var(--text-muted);
+  max-width: 520px;
+  margin-bottom: 3rem;
+  font-size: 0.95rem;
+  line-height: 1.65;
+}
+.inline-code {
+  color: var(--amber);
+  font-family: var(--font-mono);
+  font-size: 0.82em;
+}
+
+/* ===================== HOW IT WORKS ===================== */
+.steps {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(240px, 1fr));
+  gap: 1.25rem;
+}
+.step {
+  padding: 1.75rem;
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  background: var(--bg);
+  position: relative;
+  transition: border-color 0.2s ease, box-shadow 0.25s ease, transform 0.2s ease;
+}
+.step::before {
+  content: '';
+  position: absolute;
+  inset: 0;
+  border-radius: var(--radius);
+  background: linear-gradient(135deg, var(--amber-glow), transparent);
+  opacity: 0;
+  transition: opacity 0.25s ease;
+}
+.step:hover { border-color: rgba(245,158,11,0.35); transform: translateY(-3px); }
+.step:hover::before { opacity: 1; }
+.step-num {
+  font-family: var(--font-mono);
+  font-size: 2.2rem;
+  font-weight: 700;
+  color: var(--amber);
+  opacity: 0.22;
+  margin-bottom: 0.75rem;
+  line-height: 1;
+}
+.step h3 { font-size: 1rem; margin-bottom: 0.5rem; position: relative; }
+.step p  { font-size: 0.875rem; color: var(--text-muted); position: relative; }
+
+/* ===================== COMMANDS ===================== */
+.commands-section {
+  background: var(--surface);
+  border-top: 1px solid var(--border);
+  border-bottom: 1px solid var(--border);
+}
+.phase-group { margin-bottom: 2.75rem; }
+.phase-label {
+  font-family: var(--font-mono);
+  font-size: 0.67rem;
+  color: var(--amber);
+  text-transform: uppercase;
+  letter-spacing: 0.13em;
+  margin-bottom: 1rem;
+  padding-bottom: 0.5rem;
+  border-bottom: 1px solid var(--border);
+}
+.command-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fill, minmax(260px, 1fr));
+  gap: 0.875rem;
+}
 .command-card {
-  padding: 1.25rem; border: 1px solid var(--border); border-radius: 8px;
-  transition: border-color 0.2s, background 0.2s; cursor: default;
+  padding: 1.25rem 1.375rem;
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  background: var(--bg);
+  cursor: default;
+  transition: border-color 0.2s ease, box-shadow 0.25s ease, transform 0.2s ease;
+}
+.command-card:hover {
+  border-color: rgba(245,158,11,0.45);
+  box-shadow: 0 0 22px rgba(245,158,11,0.09), inset 0 0 22px rgba(245,158,11,0.02);
+  transform: translateY(-2px);
+}
+.command-card code {
+  font-family: var(--font-mono);
+  font-size: 0.74rem;
+  color: var(--amber);
+  display: block;
+  margin-bottom: 0.5rem;
+}
+.command-card p {
+  font-size: 0.82rem;
+  color: var(--text-muted);
+  line-height: 1.55;
 }
-.command-card:hover { border-color: var(--amber-dim); background: var(--surface); }
-.command-card code { font-family: var(--font-mono); font-size: 0.775rem; color: var(--amber); display: block; margin-bottom: 0.5rem; }
-.command-card p { font-size: 0.825rem; color: var(--text-muted); }
 
-/* Agents */
-.agent-grid { display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); gap: 1.5rem; }
-.agent-card { padding: 1.75rem; border: 1px solid var(--border); border-radius: 8px; }
-.agent-card h3 { font-family: var(--font-mono); color: var(--amber); margin-bottom: 0.5rem; font-size: 0.9rem; }
-.agent-card .role { font-size: 0.875rem; font-weight: 600; margin-bottom: 0.5rem; }
-.agent-card p { font-size: 0.825rem; color: var(--text-muted); }
+/* ===================== AGENTS ===================== */
+.agent-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(260px, 1fr));
+  gap: 1.25rem;
+}
+.agent-card {
+  padding: 1.75rem;
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  background: var(--bg);
+  transition: border-color 0.2s ease, box-shadow 0.25s ease, transform 0.2s ease;
+}
+.agent-card:hover {
+  border-color: rgba(245,158,11,0.4);
+  box-shadow: 0 0 36px rgba(245,158,11,0.09);
+  transform: translateY(-3px);
+}
+.agent-icon {
+  font-size: 1.4rem;
+  color: var(--amber);
+  opacity: 0.55;
+  margin-bottom: 0.75rem;
+  display: block;
+}
+.agent-card h3 {
+  font-family: var(--font-mono);
+  color: var(--amber);
+  margin-bottom: 0.4rem;
+  font-size: 0.9rem;
+}
+.agent-card .role {
+  font-size: 0.85rem;
+  font-weight: 600;
+  color: var(--text);
+  margin-bottom: 0.75rem;
+}
+.agent-card p { font-size: 0.825rem; color: var(--text-muted); line-height: 1.65; }
 
 /* v2 callout */
 .v2-callout {
-  background: linear-gradient(135deg, rgba(245,158,11,0.08), rgba(245,158,11,0.02));
-  border: 1px solid rgba(245,158,11,0.3); border-radius: 12px;
-  padding: 2.5rem; margin-top: 2rem;
+  background: linear-gradient(135deg, rgba(245,158,11,0.07), rgba(245,158,11,0.01) 60%);
+  border: 1px solid rgba(245,158,11,0.22);
+  border-radius: 12px;
+  padding: 2.5rem;
+  margin-top: 2rem;
+}
+.v2-callout h3 {
+  font-family: var(--font-mono);
+  color: var(--amber);
+  margin-bottom: 1.25rem;
+  font-size: 0.88rem;
+  letter-spacing: 0.06em;
+  text-transform: uppercase;
+}
+.v2-callout ul {
+  list-style: none;
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
+  gap: 0.625rem;
+}
+.v2-callout li {
+  font-size: 0.875rem;
+  color: var(--text-muted);
+  padding-left: 1.1rem;
+  position: relative;
+  line-height: 1.55;
+}
+.v2-callout li::before {
+  content: '→';
+  position: absolute;
+  left: 0;
+  color: var(--amber);
 }
-.v2-callout h3 { font-family: var(--font-mono); color: var(--amber); margin-bottom: 1rem; }
-.v2-callout ul { list-style: none; display: grid; grid-template-columns: repeat(auto-fit, minmax(200px, 1fr)); gap: 0.5rem; }
-.v2-callout li { font-size: 0.875rem; color: var(--text-muted); padding-left: 1rem; position: relative; }
-.v2-callout li::before { content: '→'; position: absolute; left: 0; color: var(--amber); }
 
-/* Install section */
-.install-section { background: var(--surface); border-top: 1px solid var(--border); }
+/* ===================== INSTALL ===================== */
+.install-section {
+  background: var(--surface);
+  border-top: 1px solid var(--border);
+}
 .code-block {
-  background: var(--code-bg); border: 1px solid var(--border); border-radius: 8px;
-  padding: 1.25rem 1.5rem; font-family: var(--font-mono); font-size: 0.85rem;
-  color: var(--text); margin-bottom: 1rem; white-space: pre;
+  background: var(--code-bg);
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  padding: 1.5rem 1.75rem;
+  font-family: var(--font-mono);
+  font-size: 0.85rem;
+  color: var(--text);
+  margin-bottom: 1.25rem;
+  white-space: pre;
+  overflow-x: auto;
+  line-height: 1.9;
 }
 .code-block .comment { color: var(--text-muted); }
-.code-block .cmd { color: var(--amber); }
+.code-block .cmd     { color: var(--amber); }
+.install-note { font-size: 0.875rem; color: var(--text-muted); }
 
-/* Footer */
+/* ===================== FOOTER ===================== */
 footer {
-  border-top: 1px solid var(--border); padding: 2rem;
-  text-align: center; color: var(--text-muted); font-size: 0.8rem;
+  border-top: 1px solid var(--border);
+  padding: 2rem;
+  text-align: center;
+  color: var(--text-muted);
+  font-size: 0.8rem;
 }
-footer a { color: var(--text-muted); text-decoration: none; }
+footer a { color: var(--text-muted); text-decoration: none; transition: color 0.18s ease; }
 footer a:hover { color: var(--amber); }
 
-/* Responsive */
+/* ===================== RESPONSIVE ===================== */
 @media (max-width: 640px) {
-  nav a { display: none; }
-  .install-block { flex-direction: column; gap: 0.5rem; }
+  .nav-links a { display: none; }
+  .install-block { flex-direction: column; gap: 0.5rem; text-align: center; }
+  .stats-strip { gap: 1rem; flex-wrap: wrap; }
+  .stat-num { font-size: 1.3rem; }
+  .stat-divider { height: 24px; }
+  .terminal-body { font-size: 0.75rem; }
+}
+
+/* ===================== REDUCED MOTION ===================== */
+@media (prefers-reduced-motion: reduce) {
+  *, *::before, *::after {
+    animation-duration: 0.01ms !important;
+    transition-duration: 0.01ms !important;
+  }
+  [data-animate] { opacity: 1; transform: none; }
+  .shimmer { animation: none; color: var(--amber); -webkit-text-fill-color: var(--amber); }
+  .hero-glow { animation: none; }
 }
diff --git a/scripts/detect-project.js b/scripts/detect-project.js
index 8613579..3f14844 100644
--- a/scripts/detect-project.js
+++ b/scripts/detect-project.js
@@ -43,7 +43,6 @@ try {
 
 const configDir = path.join(projectRoot, outputDir);
 const contextPath = path.join(configDir, '.qa-context.json');
-const legacyConfigPath = path.join(configDir, '.qa-config.json');
 const stateManagerPath = path.join(pluginRoot, 'scripts', 'state-manager.js');
 
 // Detection markers
@@ -762,15 +761,6 @@ const tmpContextPath = contextPath + '.tmp';
 fs.writeFileSync(tmpContextPath, JSON.stringify(projectState, null, 2));
 fs.renameSync(tmpContextPath, contextPath);
 
-// Backward compat: also write legacy .qa-config.json with detection data
-const legacyConfig = {
-    ...detection,
-    detectedAt: projectState.detectedAt,
-    projectRoot,
-    outputDir,
-};
-fs.writeFileSync(legacyConfigPath, JSON.stringify(legacyConfig, null, 2));
-
 // ---- Initialize session state ----
 
 let sessionStatus = 'new (first session)';
diff --git a/skills/shared/references/cold-start-pattern.md b/skills/shared/references/cold-start-pattern.md
index 5d1899d..0ea6530 100644
--- a/skills/shared/references/cold-start-pattern.md
+++ b/skills/shared/references/cold-start-pattern.md
@@ -14,10 +14,7 @@ outputDir = read settings.json -> outputDir (default: "qa-artifacts")
    - JSON parse error?  -> use empty defaults (do NOT error)
    - Schema mismatch?   -> use empty defaults (state-manager handles this)
 
-2. Fallback: if .qa-context.json not found, try <outputDir>/.qa-config.json
-   - Same handling: missing or corrupt -> empty defaults
-
-3. Try reading <outputDir>/.qa-session.json (if session context is relevant)
+2. Try reading <outputDir>/.qa-session.json (if session context is relevant)
    - Same handling: missing or corrupt -> empty defaults
 ```
 
@@ -96,7 +93,7 @@ If testFrameworks is empty:
 When a skill detects it ran without project context, append a footer guiding the user toward richer output.
 
 **Show the footer when:**
-- `.qa-context.json` was not found (and `.qa-config.json` fallback also missing), OR
+- `.qa-context.json` was not found, OR
 - `detection.languages` is empty or missing
 
 **Do NOT show the footer when:**
diff --git a/tests/detect-project.test.js b/tests/detect-project.test.js
index 13f2d3a..5069e2d 100644
--- a/tests/detect-project.test.js
+++ b/tests/detect-project.test.js
@@ -17,8 +17,10 @@ function runDetect(cwd) {
 }
 
 function readConfig(cwd, outputDir = 'qa-artifacts') {
-    const configPath = path.join(cwd, outputDir, '.qa-config.json');
-    return JSON.parse(fs.readFileSync(configPath, 'utf8'));
+    const contextPath = path.join(cwd, outputDir, '.qa-context.json');
+    const ctx = JSON.parse(fs.readFileSync(contextPath, 'utf8'));
+    // Flatten detection fields to the top level for backwards-compatible test assertions
+    return { ...ctx.detection, detectedAt: ctx.detectedAt, outputDir: ctx.outputDir };
 }
 
 describe('detect-project.js', () => {
@@ -38,7 +40,7 @@ describe('detect-project.js', () => {
         assert.ok(stdout.includes('No previous QA artifacts found'));
     });
 
-    it('creates .qa-config.json in output directory', () => {
+    it('creates .qa-context.json in output directory', () => {
         // Config was created by the previous test run
         const config = readConfig(tmpDir);
         assert.equal(config.outputDir, 'qa-artifacts');
@@ -189,7 +191,7 @@ describe('detect-project.js', () => {
 
         const qaDir = path.join(projectDir, 'qa-artifacts');
         const entries = fs.readdirSync(qaDir).sort();
-        // Should only have state files, no subdirectories (v2 writes .qa-context.json + legacy .qa-config.json)
+        // Should only have state files, no subdirectories (v2 writes .qa-context.json + .qa-session.json)
         const dirs = entries.filter(e => fs.statSync(path.join(qaDir, e)).isDirectory());
         assert.deepEqual(dirs, []);
     });
diff --git a/tests/hooks-toggle.test.js b/tests/hooks-toggle.test.js
index fbbeffc..edd9979 100644
--- a/tests/hooks-toggle.test.js
+++ b/tests/hooks-toggle.test.js
@@ -6,7 +6,7 @@ const { execFileSync } = require('child_process');
 
 const realPluginRoot = path.join(__dirname, '..');
 const detectScript = path.join(realPluginRoot, 'scripts', 'detect-project.js');
-const saveScript = path.join(realPluginRoot, 'scripts', 'save-artifact.js');
+const saveScript = path.join(realPluginRoot, 'scripts', 'session-hook.js');
 
 // Use a temp plugin root with a modified settings.json to avoid mutating the real one
 function makeTempPluginRoot(overrides) {
@@ -48,7 +48,7 @@ describe('hooksEnabled toggle', () => {
         assert.ok(!fs.existsSync(path.join(projectDir, 'qa-artifacts')));
     });
 
-    it('save-artifact.js exits silently when hooksEnabled is false', () => {
+    it('session-hook.js exits silently when hooksEnabled is false', () => {
         const projectDir = path.join(tmpDir, 'disabled-save');
         fs.mkdirSync(projectDir, { recursive: true });
         const fakePluginRoot = makeTempPluginRoot({ hooksEnabled: false });
@@ -83,7 +83,7 @@ describe('hooksEnabled toggle', () => {
         }
     });
 
-    it('save-artifact.js rejects absolute outputDir', () => {
+    it('session-hook.js rejects absolute outputDir', () => {
         const projectDir = path.join(tmpDir, 'abs-save');
         fs.mkdirSync(projectDir, { recursive: true });
         const fakePluginRoot = makeTempPluginRoot({ outputDir: '/tmp/evil' });
@@ -113,6 +113,6 @@ describe('hooksEnabled toggle', () => {
         });
 
         assert.ok(stdout.includes('[QA Toolkit — Project Context]'));
-        assert.ok(fs.existsSync(path.join(projectDir, 'qa-artifacts', '.qa-config.json')));
+        assert.ok(fs.existsSync(path.join(projectDir, 'qa-artifacts', '.qa-context.json')));
     });
 });

From 4b6534c2ab785b5eb58c7606e8e90ed6d45f0672 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Sat, 7 Mar 2026 05:07:50 +0530
Subject: [PATCH 11/15] =?UTF-8?q?docs:=20redesign=20landing=20page=20?=
 =?UTF-8?q?=E2=80=94=20new=20hero,=20animations,=20readability=20overhaul?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- New headline: "Test the code. Trust the release."
- Swirling particle vortex in hero, CSS background effects per section
  (aurora, light-sweep, radial-pulse, shimmer-line)
- Scroll animations: one-way reveal (removed exit animations that blanked page)
- Reordered hero: terminal above stats, bigger stat numbers with glow
- Readability: switched all body text from muted gray to white, bumped
  font sizes across cards/labels/subtitles, removed artificial max-widths
- Consistent Inter font for headings (replaced Syne display font)
- Removed unused docs/assets/ directory (19MB of unreferenced files)
- Tightened section padding, widened card grid minimums

Signed-off-by: Claude Opus 4.6 <noreply@anthropic.com>
Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 docs/index.html | 490 +++++++++++++++++++++++++++------
 docs/style.css  | 714 +++++++++++++++++++++++++++++++++++++-----------
 2 files changed, 964 insertions(+), 240 deletions(-)

diff --git a/docs/index.html b/docs/index.html
index 9b28f88..2687cf2 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -54,10 +54,21 @@
   }
   </script>
 
+  <!-- Fonts -->
+  <link rel="preconnect" href="https://fonts.googleapis.com">
+  <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
+  <link href="https://fonts.googleapis.com/css2?family=Syne:wght@600;700;800&family=Inter:wght@400;500;600&family=JetBrains+Mono:wght@400;700&display=swap" rel="stylesheet">
+
   <link rel="stylesheet" href="style.css">
 </head>
 <body>
 
+<!-- Scroll progress bar -->
+<div class="scroll-progress" id="scroll-progress"></div>
+
+<!-- Global ambient particles -->
+<canvas id="ambient-particles" aria-hidden="true"></canvas>
+
 <nav>
   <span class="logo">/qa-toolkit</span>
   <div class="nav-links">
@@ -71,14 +82,15 @@
 
 <!-- Hero -->
 <section class="hero">
-  <div class="hero-glow" aria-hidden="true"></div>
+  <div class="hero-glow" id="hero-glow" aria-hidden="true"></div>
   <div class="hero-grid" aria-hidden="true"></div>
+  <canvas id="hero-particles" aria-hidden="true"></canvas>
 
   <div class="hero-content">
     <div class="hero-badge">Claude Code Plugin &nbsp;·&nbsp; v2.0 &nbsp;·&nbsp; Free &amp; Open Source</div>
 
-    <h1>QA that <span class="shimmer">thinks for itself.</span></h1>
-    <p class="tagline">Describe what you're testing. Get a structured, stack-aware artifact — instantly. No templates. No formatting. No busywork.</p>
+    <h1>Test the code.<br><span class="shimmer">Trust the release.</span></h1>
+    <p class="tagline">15 QA commands for Claude Code — test cases, PR reviews, bug reports, and more. Auto-detects your stack. Zero config.</p>
 
     <div class="install-block">
       <span class="prompt">$</span>
@@ -86,52 +98,58 @@ <h1>QA that <span class="shimmer">thinks for itself.</span></h1>
       <button class="copy-btn" onclick="copyInstall()" aria-label="Copy install command">copy</button>
     </div>
 
+    <div class="terminal" aria-label="Terminal demo">
+      <div class="terminal-bar" aria-hidden="true">
+        <span class="dot red"></span>
+        <span class="dot yellow"></span>
+        <span class="dot green"></span>
+        <span class="terminal-title">Claude Code</span>
+      </div>
+      <div class="terminal-body" id="terminal-body"></div>
+    </div>
+
     <div class="stats-strip" aria-label="Quick stats">
       <div class="stat">
-        <span class="stat-num">15</span>
+        <span class="stat-num" data-count="15">0</span>
         <span class="stat-label">slash commands</span>
       </div>
       <div class="stat-divider" aria-hidden="true"></div>
       <div class="stat">
-        <span class="stat-num">3</span>
+        <span class="stat-num" data-count="3">0</span>
         <span class="stat-label">QA agents</span>
       </div>
       <div class="stat-divider" aria-hidden="true"></div>
       <div class="stat">
-        <span class="stat-num">0</span>
+        <span class="stat-num" data-count="0">0</span>
         <span class="stat-label">config required</span>
       </div>
       <div class="stat-divider" aria-hidden="true"></div>
       <div class="stat">
-        <span class="stat-num">8+</span>
+        <span class="stat-num" data-count-text="8+">0</span>
         <span class="stat-label">tech stacks</span>
       </div>
     </div>
 
-    <div class="terminal" aria-label="Terminal demo">
-      <div class="terminal-bar" aria-hidden="true">
-        <span class="dot red"></span>
-        <span class="dot yellow"></span>
-        <span class="dot green"></span>
-        <span class="terminal-title">Claude Code</span>
-      </div>
-      <div class="terminal-body" id="terminal-body"></div>
-    </div>
-
-    <p class="hero-hint">↓ scroll to see all 15 commands</p>
+    <p class="hero-hint">scroll to see all 15 commands</p>
   </div>
 </section>
 
 <!-- Value props -->
 <div class="quotes">
-  <div class="quotes-inner">
-    <div class="quote" data-animate>
+  <!-- Ambient orb -->
+  <div class="section-orb section-orb--amber" style="width:500px;height:500px;top:-100px;right:-150px;" data-parallax="0.03" aria-hidden="true"></div>
+  <div class="shimmer-line" style="top:0;" aria-hidden="true"></div>
+  <div class="quotes-inner" data-animate-stagger>
+    <div class="quote">
+      <div class="card-glow"></div>
       <p><strong>Bug reports in seconds.</strong> Describe what broke. Get back severity classification, steps to reproduce, expected vs actual, and environment fields — formatted for your tracker, not a generic template.</p>
     </div>
-    <div class="quote" data-animate>
+    <div class="quote">
+      <div class="card-glow"></div>
       <p><strong>PR reviews that actually help.</strong> It reads the diff, maps which files touch which user flows, and writes a QA checklist grounded in what actually changed — not a checklist you've seen a hundred times.</p>
     </div>
-    <div class="quote" data-animate>
+    <div class="quote">
+      <div class="card-glow"></div>
       <p><strong>Artifacts that fit your stack.</strong> It knows you're using Jest, not Pytest. It knows your CI runs on GitHub Actions. The output matches what your team already does, automatically.</p>
     </div>
   </div>
@@ -139,22 +157,32 @@ <h1>QA that <span class="shimmer">thinks for itself.</span></h1>
 
 <!-- How It Works -->
 <section id="how-it-works">
+  <!-- Ambient orbs -->
+  <div class="section-orb section-orb--amber-dim" style="width:400px;height:400px;top:50%;left:-100px;" data-parallax="0.04" aria-hidden="true"></div>
+  <div class="dot-grid" aria-hidden="true"></div>
+  <div class="light-sweep" aria-hidden="true"></div>
+
   <div class="container">
-    <div class="section-label">How It Works</div>
-    <h2>Three steps. That's it.</h2>
-    <p class="section-sub">Install once. Open any project. It scans your stack before you type your first command.</p>
-    <div class="steps">
-      <div class="step" data-animate>
+    <div data-animate-slide>
+      <div class="section-label">How It Works</div>
+      <h2>Three steps. That's it.</h2>
+      <p class="section-sub">Install once. Open any project. It scans your stack before you type your first command.</p>
+    </div>
+    <div class="steps" data-animate-stagger>
+      <div class="step">
+        <div class="card-glow"></div>
         <div class="step-num">01</div>
         <h3>Install the plugin</h3>
         <p>One command. Works with any project — React, Django, .NET, Go, whatever you're building.</p>
       </div>
-      <div class="step" data-animate>
+      <div class="step">
+        <div class="card-glow"></div>
         <div class="step-num">02</div>
         <h3>Open your project</h3>
         <p>On session start, QA Toolkit scans your stack and saves context. Knows your frameworks, test tools, and CI setup automatically.</p>
       </div>
-      <div class="step" data-animate>
+      <div class="step">
+        <div class="card-glow"></div>
         <div class="step-num">03</div>
         <h3>Describe what you need</h3>
         <p>Type a slash command in plain English. Because it already knows your stack, the output uses the right format, the right frameworks, the right test runner.</p>
@@ -165,47 +193,54 @@ <h3>Describe what you need</h3>
 
 <!-- Commands -->
 <section id="commands" class="commands-section">
+  <!-- Ambient orbs -->
+  <div class="section-orb section-orb--amber" style="width:600px;height:600px;bottom:-200px;right:-200px;" data-parallax="0.025" aria-hidden="true"></div>
+  <div class="aurora" aria-hidden="true"></div>
+  <div class="shimmer-line" style="top:0;" aria-hidden="true"></div>
+
   <div class="container">
-    <div class="section-label">Commands</div>
-    <h2>15 slash commands. Every QA workflow covered.</h2>
-    <p class="section-sub">Grouped by where they fit in your workflow. Each one produces a formatted artifact saved to <code class="inline-code">qa-artifacts/</code>.</p>
+    <div data-animate-slide>
+      <div class="section-label">Commands</div>
+      <h2>15 slash commands. Every QA workflow covered.</h2>
+      <p class="section-sub">Grouped by where they fit in your workflow. Each one produces a formatted artifact saved to <code class="inline-code">qa-artifacts/</code>.</p>
+    </div>
 
     <div class="phase-group" data-animate>
       <div class="phase-label">Authoring</div>
-      <div class="command-grid">
-        <div class="command-card"><code>/qa-toolkit:test-cases</code><p>Test cases from requirements — table, Gherkin, or checklist format</p></div>
-        <div class="command-card"><code>/qa-toolkit:test-plan</code><p>Full test plan with scope, strategy, and coverage targets</p></div>
-        <div class="command-card"><code>/qa-toolkit:test-data</code><p>Synthetic test data — JSON, CSV, or SQL</p></div>
-        <div class="command-card"><code>/qa-toolkit:e2e-test</code><p>Playwright test scaffold with line-by-line comments</p></div>
-        <div class="command-card"><code>/qa-toolkit:api-test</code><p>API test suite — cURL, Postman collection, or Playwright</p></div>
+      <div class="command-grid" data-animate-stagger>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:test-cases</code><p>Test cases from requirements — table, Gherkin, or checklist format</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:test-plan</code><p>Full test plan with scope, strategy, and coverage targets</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:test-data</code><p>Synthetic test data — JSON, CSV, or SQL</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:e2e-test</code><p>Playwright test scaffold with line-by-line comments</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:api-test</code><p>API test suite — cURL, Postman collection, or Playwright</p></div>
       </div>
     </div>
 
     <div class="phase-group" data-animate>
       <div class="phase-label">Review</div>
-      <div class="command-grid">
-        <div class="command-card"><code>/qa-toolkit:pr-review</code><p>Risk-flagged PR review with plain-English summary and QA checklist</p></div>
-        <div class="command-card"><code>/qa-toolkit:bug-report</code><p>Structured bug report from a casual description</p></div>
-        <div class="command-card"><code>/qa-toolkit:exploratory-testing</code><p>Exploratory test charters with session-based heuristics</p></div>
+      <div class="command-grid" data-animate-stagger>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:pr-review</code><p>Risk-flagged PR review with plain-English summary and QA checklist</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:bug-report</code><p>Structured bug report from a casual description</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:exploratory-testing</code><p>Exploratory test charters with session-based heuristics</p></div>
       </div>
     </div>
 
     <div class="phase-group" data-animate>
       <div class="phase-label">Release</div>
-      <div class="command-grid">
-        <div class="command-card"><code>/qa-toolkit:regression-planner</code><p>Risk-based regression plan with time estimates</p></div>
-        <div class="command-card"><code>/qa-toolkit:release-readiness</code><p>Go/no-go assessment with quality gate scoring</p></div>
-        <div class="command-card"><code>/qa-toolkit:risk-prioritization</code><p>Ranked risk matrix across features and change areas</p></div>
+      <div class="command-grid" data-animate-stagger>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:regression-planner</code><p>Risk-based regression plan with time estimates</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:release-readiness</code><p>Go/no-go assessment with quality gate scoring</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:risk-prioritization</code><p>Ranked risk matrix across features and change areas</p></div>
       </div>
     </div>
 
     <div class="phase-group" data-animate>
       <div class="phase-label">Analysis</div>
-      <div class="command-grid">
-        <div class="command-card"><code>/qa-toolkit:coverage-gap</code><p>Coverage gap analysis against requirements or test plan</p></div>
-        <div class="command-card"><code>/qa-toolkit:flaky-test-diagnosis</code><p>Root cause analysis for flaky tests with fix recommendations</p></div>
-        <div class="command-card"><code>/qa-toolkit:accessibility</code><p>WCAG 2.1 audit with plain-English manual test scripts</p></div>
-        <div class="command-card"><code>/qa-toolkit:setup</code><p>Read project docs, confirm detection, save preferences</p></div>
+      <div class="command-grid" data-animate-stagger>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:coverage-gap</code><p>Coverage gap analysis against requirements or test plan</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:flaky-test-diagnosis</code><p>Root cause analysis for flaky tests with fix recommendations</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:accessibility</code><p>WCAG 2.1 audit with plain-English manual test scripts</p></div>
+        <div class="command-card"><div class="card-glow"></div><code>/qa-toolkit:setup</code><p>Read project docs, confirm detection, save preferences</p></div>
       </div>
     </div>
   </div>
@@ -213,32 +248,44 @@ <h2>15 slash commands. Every QA workflow covered.</h2>
 
 <!-- Agents -->
 <section id="agents">
+  <!-- Ambient orbs -->
+  <div class="section-orb section-orb--amber" style="width:500px;height:500px;top:20%;left:-150px;" data-parallax="0.035" aria-hidden="true"></div>
+  <div class="section-orb section-orb--amber-dim" style="width:350px;height:350px;bottom:10%;right:-100px;" data-parallax="0.05" aria-hidden="true"></div>
+  <div class="dot-grid" aria-hidden="true"></div>
+  <div class="radial-pulse" style="top:50%;left:50%;margin-left:-250px;margin-top:-250px;" aria-hidden="true"></div>
+  <div class="shimmer-line" style="bottom:0;" aria-hidden="true"></div>
+
   <div class="container">
-    <div class="section-label">Agents</div>
-    <h2>When you need a conversation, not a document.</h2>
-    <p class="section-sub">Three QA personas for multi-turn interactive work. They remember context across the conversation and adapt as you share more.</p>
-    <div class="agent-grid">
-      <div class="agent-card" data-animate>
-        <div class="agent-icon" aria-hidden="true">◈</div>
+    <div data-animate-slide>
+      <div class="section-label">Agents</div>
+      <h2>When you need a conversation, not a document.</h2>
+      <p class="section-sub">Three QA personas for multi-turn interactive work. They remember context across the conversation and adapt as you share more.</p>
+    </div>
+    <div class="agent-grid" data-animate-stagger>
+      <div class="agent-card">
+        <div class="card-glow"></div>
+        <div class="agent-icon" aria-hidden="true">&#9672;</div>
         <h3>qa-reviewer</h3>
         <p class="role">The Code Reviewer's QA Brain</p>
         <p>Walk me through this PR. What should I actually be testing? Translates code changes into testing impact in plain English.</p>
       </div>
-      <div class="agent-card" data-animate>
-        <div class="agent-icon" aria-hidden="true">◈</div>
+      <div class="agent-card">
+        <div class="card-glow"></div>
+        <div class="agent-icon" aria-hidden="true">&#9672;</div>
         <h3>qa-explorer</h3>
         <p class="role">The Edge Case Hunter</p>
         <p>Help me find what could break in this feature. Generates exploratory test charters and surfaces edge cases you wouldn't think to test.</p>
       </div>
-      <div class="agent-card" data-animate>
-        <div class="agent-icon" aria-hidden="true">◈</div>
+      <div class="agent-card">
+        <div class="card-glow"></div>
+        <div class="agent-icon" aria-hidden="true">&#9672;</div>
         <h3>qa-lead</h3>
         <p class="role">The Release Decision Maker</p>
         <p>Let's decide if we're ready to ship. Makes release decisions, plans regression scope, and produces executive summaries.</p>
       </div>
     </div>
 
-    <div class="v2-callout" data-animate>
+    <div class="v2-callout" data-animate-scale>
       <h3>What's New in v2.0</h3>
       <ul>
         <li>State-aware skills — every command reads your detected project context</li>
@@ -253,16 +300,21 @@ <h3>What's New in v2.0</h3>
 
 <!-- Install -->
 <section id="install" class="install-section">
+  <div class="section-orb section-orb--amber" style="width:400px;height:400px;top:50%;left:50%;transform:translate(-50%,-50%);" data-parallax="0.02" aria-hidden="true"></div>
+  <div class="light-sweep" aria-hidden="true"></div>
+
   <div class="container">
-    <div class="section-label">Install</div>
-    <h2>One command to install. Zero to configure.</h2>
-    <p class="section-sub">Works with any project. Configurable if you want — but you don't have to.</p>
-    <div class="code-block" data-animate><span class="comment"># From GitHub</span>
+    <div data-animate-slide>
+      <div class="section-label">Install</div>
+      <h2>One command to install. Zero to configure.</h2>
+      <p class="section-sub">Works with any project. Configurable if you want — but you don't have to.</p>
+    </div>
+    <div class="code-block" data-animate-scale><span class="comment"># From GitHub</span>
 <span class="cmd">claude plugin add github:cyberwalk3r/qa-toolkit</span>
 
 <span class="comment"># Local development</span>
 <span class="cmd">claude plugin add ./qa-toolkit</span></div>
-    <p class="install-note">On first session, auto-detection runs and saves your project context to <code class="inline-code">qa-artifacts/.qa-context.json</code>. Every subsequent command reads from it.</p>
+    <p class="install-note" data-animate>On first session, auto-detection runs and saves your project context to <code class="inline-code">qa-artifacts/.qa-context.json</code>. Every subsequent command reads from it.</p>
   </div>
 </section>
 
@@ -282,42 +334,260 @@ <h2>One command to install. Zero to configure.</h2>
   });
 }
 
+// ---- Scroll progress bar ----
+const scrollProgress = document.getElementById('scroll-progress');
+function updateScrollProgress() {
+  const scrollTop = window.scrollY;
+  const docHeight = document.documentElement.scrollHeight - window.innerHeight;
+  const progress = docHeight > 0 ? (scrollTop / docHeight) * 100 : 0;
+  scrollProgress.style.width = progress + '%';
+}
+
+// ---- Nav scroll effect ----
+const nav = document.querySelector('nav');
+window.addEventListener('scroll', () => {
+  nav.classList.toggle('scrolled', window.scrollY > 40);
+  updateScrollProgress();
+}, { passive: true });
+
+// ---- Mouse-reactive hero glow ----
+const heroGlow = document.getElementById('hero-glow');
+const heroSection = document.querySelector('.hero');
+heroSection.addEventListener('mousemove', (e) => {
+  const rect = heroSection.getBoundingClientRect();
+  const x = e.clientX - rect.left;
+  const y = e.clientY - rect.top;
+  heroGlow.style.transform = `translate(${x - 350}px, ${y - 350}px)`;
+});
+
+// ---- Global ambient particles (all sections) ----
+(function initAmbientParticles() {
+  const canvas = document.getElementById('ambient-particles');
+  if (!canvas) return;
+  const ctx = canvas.getContext('2d');
+  const dpr = Math.min(window.devicePixelRatio || 1, 2);
+  let particles = [];
+  const COUNT = 60;
+
+  function resize() {
+    canvas.width = window.innerWidth * dpr;
+    canvas.height = window.innerHeight * dpr;
+    canvas.style.width = window.innerWidth + 'px';
+    canvas.style.height = window.innerHeight + 'px';
+    ctx.setTransform(dpr, 0, 0, dpr, 0, 0);
+  }
+
+  function seed() {
+    particles = [];
+    const w = window.innerWidth;
+    const h = window.innerHeight;
+    for (let i = 0; i < COUNT; i++) {
+      particles.push({
+        x: Math.random() * w,
+        y: Math.random() * h,
+        r: Math.random() * 1.2 + 0.3,
+        vx: (Math.random() - 0.5) * 0.12,
+        vy: -Math.random() * 0.25 - 0.03,
+        alpha: Math.random() * 0.35 + 0.05,
+        pulse: Math.random() * Math.PI * 2,
+        pulseSpeed: Math.random() * 0.008 + 0.003,
+      });
+    }
+  }
+
+  function draw() {
+    const w = window.innerWidth;
+    const h = window.innerHeight;
+    ctx.clearRect(0, 0, w, h);
+
+    for (const p of particles) {
+      p.x += p.vx;
+      p.y += p.vy;
+      p.pulse += p.pulseSpeed;
+
+      if (p.y < -10) { p.y = h + 10; p.x = Math.random() * w; }
+      if (p.x < -10) p.x = w + 10;
+      if (p.x > w + 10) p.x = -10;
+
+      const flicker = 0.5 + 0.5 * Math.sin(p.pulse);
+      const a = p.alpha * flicker;
+      ctx.beginPath();
+      ctx.arc(p.x, p.y, p.r, 0, Math.PI * 2);
+      ctx.fillStyle = `rgba(245, 158, 11, ${a})`;
+      ctx.fill();
+
+      if (p.r > 0.8) {
+        ctx.beginPath();
+        ctx.arc(p.x, p.y, p.r * 3, 0, Math.PI * 2);
+        ctx.fillStyle = `rgba(245, 158, 11, ${a * 0.1})`;
+        ctx.fill();
+      }
+    }
+    requestAnimationFrame(draw);
+  }
+
+  resize();
+  seed();
+  draw();
+  window.addEventListener('resize', () => { resize(); seed(); });
+})();
+
+// ---- Hero swirling particle vortex ----
+(function initHeroVortex() {
+  const canvas = document.getElementById('hero-particles');
+  if (!canvas) return;
+  const ctx = canvas.getContext('2d');
+  const dpr = Math.min(window.devicePixelRatio || 1, 2);
+  let particles = [];
+  const COUNT = 350;
+  let time = 0;
+
+  function resize() {
+    const rect = heroSection.getBoundingClientRect();
+    canvas.width = rect.width * dpr;
+    canvas.height = rect.height * dpr;
+    canvas.style.width = rect.width + 'px';
+    canvas.style.height = rect.height + 'px';
+    ctx.setTransform(dpr, 0, 0, dpr, 0, 0);
+  }
+
+  function seed() {
+    particles = [];
+    const w = canvas.width / dpr;
+    const h = canvas.height / dpr;
+    const maxDist = Math.max(w, h) * 0.55;
+    for (let i = 0; i < COUNT; i++) {
+      const angle = Math.random() * Math.PI * 2;
+      const dist = 30 + Math.random() * maxDist;
+      particles.push({
+        angle: angle,
+        dist: dist,
+        baseDist: dist,
+        speed: (0.001 + Math.random() * 0.004) * (Math.random() < 0.5 ? 1 : 0.6),
+        r: Math.random() * 1.4 + 0.2,
+        alpha: Math.random() * 0.7 + 0.15,
+        pulse: Math.random() * Math.PI * 2,
+        pulseSpeed: Math.random() * 0.012 + 0.004,
+        wobble: Math.random() * 25,
+        wobbleSpeed: Math.random() * 0.015 + 0.008,
+      });
+    }
+  }
+
+  function draw() {
+    const w = canvas.width / dpr;
+    const h = canvas.height / dpr;
+    const cx = w / 2;
+    const cy = h / 2;
+    time++;
+
+    // Soft fade — leaves motion trails
+    ctx.fillStyle = 'rgba(10, 10, 10, 0.18)';
+    ctx.fillRect(0, 0, w, h);
+
+    for (const p of particles) {
+      p.angle += p.speed;
+      p.pulse += p.pulseSpeed;
+      const wobbleOffset = Math.sin(time * p.wobbleSpeed) * p.wobble;
+      const dist = p.baseDist + wobbleOffset;
+
+      const x = cx + Math.cos(p.angle) * dist;
+      const y = cy + Math.sin(p.angle) * dist * 0.6;
+
+      const flicker = 0.5 + 0.5 * Math.sin(p.pulse);
+      const a = p.alpha * flicker;
+
+      // Main particle
+      ctx.beginPath();
+      ctx.arc(x, y, p.r, 0, Math.PI * 2);
+      ctx.fillStyle = `rgba(245, 158, 11, ${a})`;
+      ctx.fill();
+
+      // Soft glow
+      if (p.r > 0.7) {
+        ctx.beginPath();
+        ctx.arc(x, y, p.r * 3, 0, Math.PI * 2);
+        ctx.fillStyle = `rgba(245, 158, 11, ${a * 0.12})`;
+        ctx.fill();
+      }
+    }
+    requestAnimationFrame(draw);
+  }
+
+  resize();
+  seed();
+  draw();
+  window.addEventListener('resize', () => { resize(); seed(); });
+})();
+
+// ---- Card glow tracking ----
+document.addEventListener('mousemove', (e) => {
+  const cards = document.querySelectorAll('.quote, .step, .command-card, .agent-card');
+  for (const card of cards) {
+    const glow = card.querySelector('.card-glow');
+    if (!glow) continue;
+    const rect = card.getBoundingClientRect();
+    const x = e.clientX - rect.left;
+    const y = e.clientY - rect.top;
+    glow.style.left = x + 'px';
+    glow.style.top = y + 'px';
+  }
+});
+
+// ---- Parallax orbs on scroll ----
+const parallaxOrbs = document.querySelectorAll('[data-parallax]');
+let ticking = false;
+window.addEventListener('scroll', () => {
+  if (!ticking) {
+    requestAnimationFrame(() => {
+      const scrollY = window.scrollY;
+      parallaxOrbs.forEach(orb => {
+        const speed = parseFloat(orb.dataset.parallax) || 0.03;
+        const offset = scrollY * speed;
+        orb.style.transform = `translateY(${-offset}px)`;
+      });
+      ticking = false;
+    });
+    ticking = true;
+  }
+}, { passive: true });
+
 // ---- Terminal cycling demos ----
 const demos = [
   {
     cmd: '> /qa-toolkit:bug-report The checkout page crashes with long addresses',
     lines: [
       { cls: 'line-out', text: 'Detecting project context...' },
-      { cls: 'line-key', text: 'Stack: <span class="line-val">React · Node.js · Jest · GitHub Actions</span>' },
+      { cls: 'line-key', text: 'Stack: <span class="line-val">React \u00b7 Node.js \u00b7 Jest \u00b7 GitHub Actions</span>' },
       { cls: 'line-key', text: 'Tracker: <span class="line-val">GitHub Issues</span>' },
       { cls: 'line-spacer', text: '' },
       { cls: 'line-key', text: 'Title: <span class="line-val">Checkout crashes on addresses &gt; 60 chars [P1]</span>' },
       { cls: 'line-key', text: 'Actual: <span class="line-val">TypeError: Cannot read property \'length\'</span>' },
-      { cls: 'line-out', text: 'Saved → qa-artifacts/bug-reports/2026-03-06-checkout-crash.md' },
+      { cls: 'line-out', text: 'Saved \u2192 qa-artifacts/bug-reports/2026-03-06-checkout-crash.md' },
     ]
   },
   {
     cmd: '> /qa-toolkit:pr-review feature/new-auth-flow',
     lines: [
-      { cls: 'line-out', text: 'Reading diff · 12 files changed...' },
-      { cls: 'line-key', text: 'Risk Level: <span class="line-val line-amber">HIGH — auth flow changed</span>' },
-      { cls: 'line-key', text: 'Affected flows: <span class="line-val">login · session · password-reset</span>' },
+      { cls: 'line-out', text: 'Reading diff \u00b7 12 files changed...' },
+      { cls: 'line-key', text: 'Risk Level: <span class="line-val line-amber">HIGH \u2014 auth flow changed</span>' },
+      { cls: 'line-key', text: 'Affected flows: <span class="line-val">login \u00b7 session \u00b7 password-reset</span>' },
       { cls: 'line-spacer', text: '' },
       { cls: 'line-key', text: 'QA Checklist: <span class="line-val">7 items generated</span>' },
-      { cls: 'line-out', text: '  ✓ Verify session expiry after token rotation' },
-      { cls: 'line-out', text: 'Saved → qa-artifacts/pr-reviews/pr-review-2026-03-06-auth-flow.md' },
+      { cls: 'line-out', text: '  \u2713 Verify session expiry after token rotation' },
+      { cls: 'line-out', text: 'Saved \u2192 qa-artifacts/pr-reviews/pr-review-2026-03-06-auth-flow.md' },
     ]
   },
   {
     cmd: '> /qa-toolkit:test-cases User registration with email verification',
     lines: [
       { cls: 'line-out', text: 'Reading project context...' },
-      { cls: 'line-key', text: 'Framework: <span class="line-val">Playwright detected — using BDD format</span>' },
+      { cls: 'line-key', text: 'Framework: <span class="line-val">Playwright detected \u2014 using BDD format</span>' },
       { cls: 'line-spacer', text: '' },
-      { cls: 'line-key', text: 'TC-REG-01 <span class="line-val">P0 · Happy path registration</span>' },
-      { cls: 'line-key', text: 'TC-REG-02 <span class="line-val">P0 · Duplicate email rejected</span>' },
-      { cls: 'line-key', text: 'TC-REG-03 <span class="line-val">P1 · Verification link expires</span>' },
-      { cls: 'line-out', text: 'Saved → qa-artifacts/test-cases/test-cases-2026-03-06-registration.md' },
+      { cls: 'line-key', text: 'TC-REG-01 <span class="line-val">P0 \u00b7 Happy path registration</span>' },
+      { cls: 'line-key', text: 'TC-REG-02 <span class="line-val">P0 \u00b7 Duplicate email rejected</span>' },
+      { cls: 'line-key', text: 'TC-REG-03 <span class="line-val">P1 \u00b7 Verification link expires</span>' },
+      { cls: 'line-out', text: 'Saved \u2192 qa-artifacts/test-cases/test-cases-2026-03-06-registration.md' },
     ]
   }
 ];
@@ -341,9 +611,13 @@ <h2>One command to install. Zero to configure.</h2>
     div.className = line.cls || 'line-spacer';
     div.innerHTML = line.text;
     div.style.opacity = '0';
-    div.style.transition = 'opacity 0.35s ease';
+    div.style.transform = 'translateY(4px)';
+    div.style.transition = 'opacity 0.4s ease, transform 0.4s ease';
     terminalBody.appendChild(div);
-    setTimeout(() => { div.style.opacity = '1'; }, revealDelay + i * 200);
+    setTimeout(() => {
+      div.style.opacity = '1';
+      div.style.transform = 'translateY(0)';
+    }, revealDelay + i * 220);
   });
 
   const cursor = document.createElement('span');
@@ -363,19 +637,59 @@ <h2>One command to install. Zero to configure.</h2>
 renderDemo(0, true);
 setInterval(cycleDemo, 7000);
 
-// ---- Scroll reveal ----
-const observer = new IntersectionObserver((entries) => {
+// ---- Stat count-up ----
+function animateStats() {
+  document.querySelectorAll('.stat-num[data-count]').forEach(el => {
+    const target = parseInt(el.dataset.count, 10);
+    if (target === 0) { el.textContent = '0'; return; }
+    let current = 0;
+    const step = Math.max(1, Math.floor(target / 20));
+    const interval = setInterval(() => {
+      current += step;
+      if (current >= target) { current = target; clearInterval(interval); }
+      el.textContent = current;
+    }, 40);
+  });
+  document.querySelectorAll('.stat-num[data-count-text]').forEach(el => {
+    const text = el.dataset.countText;
+    const num = parseInt(text, 10);
+    let current = 0;
+    const interval = setInterval(() => {
+      current++;
+      if (current >= num) { current = num; clearInterval(interval); el.textContent = text; }
+      else { el.textContent = current; }
+    }, 60);
+  });
+}
+
+const statsStrip = document.querySelector('.stats-strip');
+let statsCounted = false;
+const statsObserver = new IntersectionObserver((entries) => {
+  if (entries[0].isIntersecting && !statsCounted) {
+    statsCounted = true;
+    setTimeout(animateStats, 200);
+    statsObserver.disconnect();
+  }
+}, { threshold: 0.5 });
+statsObserver.observe(statsStrip);
+
+// ---- Scroll reveal (one-way — stays visible once shown) ----
+const animSelectors = '[data-animate], [data-animate-stagger], [data-animate-slide], [data-animate-scale]';
+
+const scrollObserver = new IntersectionObserver((entries) => {
   entries.forEach(entry => {
     if (entry.isIntersecting) {
       entry.target.classList.add('visible');
-      observer.unobserve(entry.target);
+      scrollObserver.unobserve(entry.target);
     }
   });
-}, { threshold: 0.08 });
+}, {
+  threshold: 0.1,
+  rootMargin: '0px 0px -60px 0px'
+});
 
-document.querySelectorAll('[data-animate]').forEach((el, i) => {
-  el.style.transitionDelay = (i % 4 * 0.1) + 's';
-  observer.observe(el);
+document.querySelectorAll(animSelectors).forEach(el => {
+  scrollObserver.observe(el);
 });
 </script>
 
diff --git a/docs/style.css b/docs/style.css
index 57c6685..c30abf5 100644
--- a/docs/style.css
+++ b/docs/style.css
@@ -2,23 +2,26 @@
 *, *::before, *::after { box-sizing: border-box; margin: 0; padding: 0; }
 
 :root {
-  --bg: #0d0d0d;
-  --surface: #141414;
-  --surface-2: #1a1a1a;
-  --border: #252525;
+  --bg: #0a0a0a;
+  --surface: #111111;
+  --surface-2: #171717;
+  --border: #1e1e1e;
   --border-hover: rgba(245,158,11,0.45);
   --amber: #f59e0b;
   --amber-dim: #d97706;
   --amber-glow: rgba(245,158,11,0.12);
+  --amber-faint: rgba(245,158,11,0.04);
   --green: #34d399;
   --text: #e5e7eb;
   --text-muted: #6b7280;
-  --code-bg: #111111;
+  --code-bg: #0e0e0e;
+  --font-display: 'Syne', 'Inter', sans-serif;
   --font-mono: 'JetBrains Mono', 'Fira Code', 'Cascadia Code', monospace;
-  --font-sans: -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;
+  --font-sans: 'Inter', -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;
   --radius: 10px;
   --radius-sm: 6px;
   --ease: cubic-bezier(0.22, 1, 0.36, 1);
+  --ease-out-expo: cubic-bezier(0.16, 1, 0.3, 1);
 }
 
 html { scroll-behavior: smooth; }
@@ -30,15 +33,41 @@ body {
   overflow-x: hidden;
 }
 
+/* ===================== GRAIN OVERLAY ===================== */
+body::after {
+  content: '';
+  position: fixed;
+  inset: 0;
+  z-index: 9999;
+  pointer-events: none;
+  opacity: 0.028;
+  background-image: url("data:image/svg+xml,%3Csvg viewBox='0 0 256 256' xmlns='http://www.w3.org/2000/svg'%3E%3Cfilter id='n'%3E%3CfeTurbulence type='fractalNoise' baseFrequency='0.9' numOctaves='4' stitchTiles='stitch'/%3E%3C/filter%3E%3Crect width='100%25' height='100%25' filter='url(%23n)'/%3E%3C/svg%3E");
+  background-size: 180px;
+}
+
+/* ===================== GLOBAL AMBIENT CANVAS ===================== */
+#ambient-particles {
+  position: fixed;
+  inset: 0;
+  pointer-events: none;
+  z-index: 0;
+  opacity: 0.85;
+}
+
 /* ===================== NAV ===================== */
 nav {
   position: fixed; top: 0; left: 0; right: 0; z-index: 100;
-  background: rgba(13,13,13,0.82);
-  backdrop-filter: blur(16px) saturate(180%);
-  -webkit-backdrop-filter: blur(16px) saturate(180%);
-  border-bottom: 1px solid var(--border);
+  background: rgba(10,10,10,0.7);
+  backdrop-filter: blur(20px) saturate(180%);
+  -webkit-backdrop-filter: blur(20px) saturate(180%);
+  border-bottom: 1px solid transparent;
   padding: 0.75rem 2rem;
   display: flex; align-items: center; justify-content: space-between;
+  transition: border-color 0.4s ease, box-shadow 0.4s ease;
+}
+nav.scrolled {
+  border-bottom-color: var(--border);
+  box-shadow: 0 1px 0 0 rgba(245,158,11,0.08), 0 8px 32px rgba(0,0,0,0.4);
 }
 nav .logo {
   font-family: var(--font-mono);
@@ -50,11 +79,140 @@ nav .logo {
 .nav-links a {
   color: var(--text-muted);
   text-decoration: none;
-  font-size: 0.875rem;
+  font-size: 0.84rem;
   margin-left: 2rem;
   transition: color 0.18s ease;
+  position: relative;
+}
+.nav-links a::after {
+  content: '';
+  position: absolute;
+  bottom: -4px;
+  left: 0;
+  width: 0%;
+  height: 1px;
+  background: var(--amber);
+  transition: width 0.3s var(--ease);
+}
+.nav-links a:hover { color: var(--text); }
+.nav-links a:hover::after { width: 100%; }
+
+/* ===================== SECTION TRANSITIONS ===================== */
+/* Gradient bleed between sections */
+.section-transition {
+  position: absolute;
+  left: 0; right: 0;
+  height: 120px;
+  pointer-events: none;
+  z-index: 1;
+}
+.section-transition--top {
+  top: -60px;
+  background: linear-gradient(180deg, transparent, var(--bg));
+}
+.section-transition--top-surface {
+  top: -60px;
+  background: linear-gradient(180deg, transparent, var(--surface));
+}
+.section-transition--bottom {
+  bottom: -60px;
+  background: linear-gradient(180deg, var(--bg), transparent);
+}
+.section-transition--bottom-surface {
+  bottom: -60px;
+  background: linear-gradient(180deg, var(--surface), transparent);
+}
+
+/* Section ambient glow orbs — parallax via JS */
+.section-orb {
+  position: absolute;
+  border-radius: 50%;
+  pointer-events: none;
+  filter: blur(80px);
+  will-change: transform;
+  transition: transform 0.1s linear;
+}
+.section-orb--amber {
+  background: radial-gradient(circle, rgba(245,158,11,0.18), transparent 70%);
+}
+.section-orb--amber-dim {
+  background: radial-gradient(circle, rgba(217,119,6,0.12), transparent 70%);
+}
+
+/* ===================== AURORA / GRADIENT SWEEP ===================== */
+/* Slow-moving gradient wash — used on select sections */
+.aurora {
+  position: absolute;
+  inset: 0;
+  pointer-events: none;
+  opacity: 0.7;
+  background:
+    radial-gradient(ellipse 60% 50% at 20% 60%, rgba(245,158,11,0.14), transparent 70%),
+    radial-gradient(ellipse 50% 40% at 80% 30%, rgba(217,119,6,0.10), transparent 70%);
+  animation: aurora-drift 18s ease-in-out infinite alternate;
+  will-change: transform;
+}
+@keyframes aurora-drift {
+  0%   { transform: translate(0, 0) scale(1); }
+  100% { transform: translate(-30px, 20px) scale(1.05); }
+}
+
+/* Horizontal light sweep — subtle scan effect */
+.light-sweep {
+  position: absolute;
+  inset: 0;
+  pointer-events: none;
+  overflow: hidden;
+}
+.light-sweep::after {
+  content: '';
+  position: absolute;
+  top: 0;
+  left: -100%;
+  width: 60%;
+  height: 100%;
+  background: linear-gradient(90deg, transparent, rgba(245,158,11,0.07), transparent);
+  animation: sweep 12s ease-in-out infinite;
+  will-change: transform;
+}
+@keyframes sweep {
+  0%   { transform: translateX(0); }
+  50%  { transform: translateX(400%); }
+  100% { transform: translateX(0); }
+}
+
+/* Radial pulse — gentle breathing glow */
+.radial-pulse {
+  position: absolute;
+  width: 500px;
+  height: 500px;
+  border-radius: 50%;
+  pointer-events: none;
+  background: radial-gradient(circle, rgba(245,158,11,0.14), transparent 70%);
+  animation: pulse-breathe 8s ease-in-out infinite;
+  will-change: transform, opacity;
+}
+@keyframes pulse-breathe {
+  0%, 100% { transform: scale(1); opacity: 0.5; }
+  50%      { transform: scale(1.15); opacity: 0.8; }
+}
+
+/* Noise shimmer band — thin animated gradient line */
+.shimmer-line {
+  position: absolute;
+  left: 0;
+  right: 0;
+  height: 1px;
+  pointer-events: none;
+  background: linear-gradient(90deg, transparent 0%, rgba(245,158,11,0.35) 30%, rgba(245,158,11,0.6) 50%, rgba(245,158,11,0.35) 70%, transparent 100%);
+  background-size: 200% 100%;
+  animation: shimmer-band 4s ease-in-out infinite;
+  opacity: 0.8;
+}
+@keyframes shimmer-band {
+  0%   { background-position: -200% 0; }
+  100% { background-position: 200% 0; }
 }
-.nav-links a:hover { color: var(--amber); }
 
 /* ===================== HERO ===================== */
 .hero {
@@ -69,90 +227,111 @@ nav .logo {
   text-align: center;
 }
 
-/* Dot grid background */
-.hero-grid {
+/* Dot grid background — shared class for multiple sections */
+.dot-grid {
+  position: absolute; inset: 0;
+  background-image: radial-gradient(circle at 1px 1px, rgba(245,158,11,0.15) 1px, transparent 0);
+  background-size: 32px 32px;
+  pointer-events: none;
+  mask-image: radial-gradient(ellipse 80% 80% at 50% 50%, black 20%, transparent 100%);
+  -webkit-mask-image: radial-gradient(ellipse 80% 80% at 50% 50%, black 20%, transparent 100%);
+}
+.hero-grid { composes: dot-grid; }
+.hero-grid,
+.dot-grid {
   position: absolute; inset: 0;
-  background-image: radial-gradient(circle at 1px 1px, rgba(245,158,11,0.13) 1px, transparent 0);
+  background-image: radial-gradient(circle at 1px 1px, rgba(245,158,11,0.15) 1px, transparent 0);
   background-size: 32px 32px;
   pointer-events: none;
-  mask-image: radial-gradient(ellipse 85% 85% at 50% 50%, black 30%, transparent 100%);
-  -webkit-mask-image: radial-gradient(ellipse 85% 85% at 50% 50%, black 30%, transparent 100%);
+  mask-image: radial-gradient(ellipse 80% 80% at 50% 50%, black 20%, transparent 100%);
+  -webkit-mask-image: radial-gradient(ellipse 80% 80% at 50% 50%, black 20%, transparent 100%);
 }
 
-/* Ambient amber glow */
+/* Mouse-reactive ambient glow — positioned via JS */
 .hero-glow {
   position: absolute;
-  width: 900px; height: 900px;
-  background: radial-gradient(circle, rgba(245,158,11,0.09) 0%, transparent 65%);
+  width: 700px; height: 700px;
+  background: radial-gradient(circle, rgba(245,158,11,0.18) 0%, rgba(245,158,11,0.06) 40%, transparent 70%);
   top: 50%; left: 50%;
   transform: translate(-50%, -50%);
   pointer-events: none;
-  animation: pulse-glow 7s ease-in-out infinite;
+  transition: transform 0.15s linear, opacity 0.4s ease;
+  will-change: transform;
 }
-@keyframes pulse-glow {
-  0%, 100% { opacity: 0.6; transform: translate(-50%, -50%) scale(1); }
-  50% { opacity: 1; transform: translate(-50%, -50%) scale(1.1); }
+
+/* Particle canvas */
+#hero-particles {
+  position: absolute;
+  inset: 0;
+  pointer-events: none;
+  z-index: 0;
+  opacity: 0.9;
 }
 
 .hero-content {
   position: relative;
   z-index: 1;
   width: 100%;
-  max-width: 820px;
+  max-width: 960px;
 }
 
 /* Badge */
 .hero-badge {
   display: inline-block;
   font-family: var(--font-mono);
-  font-size: 0.7rem;
+  font-size: 0.74rem;
   color: var(--amber);
-  border: 1px solid rgba(245,158,11,0.3);
-  background: rgba(245,158,11,0.06);
+  border: 1px solid rgba(245,158,11,0.35);
+  background: rgba(245,158,11,0.08);
   border-radius: 100px;
   padding: 0.3rem 1rem;
-  margin-bottom: 1.5rem;
+  margin-bottom: 1.75rem;
   letter-spacing: 0.06em;
-  animation: fade-up 0.6s var(--ease) both;
+  opacity: 0;
+  animation: hero-enter 1s var(--ease-out-expo) 0.2s forwards;
 }
 
 .hero h1 {
-  font-family: var(--font-mono);
-  font-size: clamp(2.1rem, 5.5vw, 4rem);
-  font-weight: 700;
-  letter-spacing: -0.03em;
-  line-height: 1.07;
-  animation: fade-up 0.65s var(--ease) 0.1s both;
+  font-family: var(--font-sans);
+  font-size: clamp(2.2rem, 5.5vw, 3.6rem);
+  font-weight: 600;
+  letter-spacing: -0.02em;
+  line-height: 1.2;
+  opacity: 0;
+  animation: hero-enter-blur 1.2s var(--ease-out-expo) 0.35s forwards;
 }
 
 /* Shimmer gradient text */
 .shimmer {
   background: linear-gradient(
-    90deg,
-    #f59e0b 0%,
-    #fde68a 25%,
-    #f59e0b 45%,
-    #d97706 65%,
-    #fbbf24 80%,
-    #f59e0b 100%
+    105deg,
+    #d97706 0%,
+    #fbbf24 18%,
+    #fde68a 36%,
+    #f59e0b 54%,
+    #d97706 72%,
+    #fbbf24 90%,
+    #d97706 100%
   );
-  background-size: 300% auto;
+  background-size: 250% auto;
   -webkit-background-clip: text;
   -webkit-text-fill-color: transparent;
   background-clip: text;
-  animation: shimmer 5s linear infinite;
+  animation: shimmer 6s ease-in-out infinite;
 }
 @keyframes shimmer {
   0%   { background-position: 0% center; }
-  100% { background-position: 300% center; }
+  50%  { background-position: 250% center; }
+  100% { background-position: 0% center; }
 }
 
 .hero .tagline {
-  font-size: clamp(0.95rem, 2vw, 1.2rem);
-  color: var(--text-muted);
-  margin: 1.25rem auto 2rem;
-  max-width: 520px;
-  animation: fade-up 0.65s var(--ease) 0.18s both;
+  font-size: clamp(1rem, 2.2vw, 1.2rem);
+  color: var(--text);
+  margin: 1.5rem auto 2.25rem;
+  max-width: 700px;
+  opacity: 0;
+  animation: hero-enter 1s var(--ease-out-expo) 0.55s forwards;
 }
 
 /* Install block */
@@ -165,14 +344,15 @@ nav .logo {
   border-radius: var(--radius);
   padding: 0.875rem 1.5rem;
   font-family: var(--font-mono);
-  font-size: 0.875rem;
-  margin-bottom: 2rem;
-  animation: fade-up 0.65s var(--ease) 0.26s both;
-  transition: border-color 0.2s ease, box-shadow 0.2s ease;
+  font-size: 0.9rem;
+  margin-bottom: 2.25rem;
+  opacity: 0;
+  animation: hero-enter 1s var(--ease-out-expo) 0.7s forwards;
+  transition: border-color 0.25s ease, box-shadow 0.25s ease;
 }
 .install-block:hover {
-  border-color: var(--border-hover);
-  box-shadow: 0 0 28px rgba(245,158,11,0.1);
+  border-color: rgba(245,158,11,0.35);
+  box-shadow: 0 0 40px rgba(245,158,11,0.08), 0 0 80px rgba(245,158,11,0.04);
 }
 .install-block .prompt { color: var(--amber); }
 
@@ -185,42 +365,47 @@ nav .logo {
   padding: 0.25rem 0.7rem;
   font-family: var(--font-mono);
   font-size: 0.7rem;
-  transition: border-color 0.18s ease, color 0.18s ease;
+  transition: border-color 0.18s ease, color 0.18s ease, transform 0.15s ease;
   white-space: nowrap;
 }
-.copy-btn:hover,
-.copy-btn.copied { border-color: var(--amber); color: var(--amber); }
+.copy-btn:hover { border-color: var(--amber); color: var(--amber); transform: scale(1.04); }
+.copy-btn.copied { border-color: var(--green); color: var(--green); }
 
 /* Stats strip */
 .stats-strip {
   display: flex;
   align-items: center;
   justify-content: center;
-  gap: 1.75rem;
-  margin-bottom: 2.5rem;
-  animation: fade-up 0.65s var(--ease) 0.34s both;
+  gap: 2.5rem;
+  margin-top: 2.5rem;
+  margin-bottom: 1.5rem;
+  opacity: 0;
+  animation: hero-enter 1s var(--ease-out-expo) 1.4s forwards;
 }
 .stat { text-align: center; }
 .stat-num {
   display: block;
   font-family: var(--font-mono);
-  font-size: 1.6rem;
+  font-size: 2.4rem;
   font-weight: 700;
   color: var(--amber);
   line-height: 1;
+  font-variant-numeric: tabular-nums;
+  text-shadow: 0 0 20px rgba(245,158,11,0.4);
 }
 .stat-label {
   display: block;
-  font-size: 0.68rem;
-  color: var(--text-muted);
+  font-size: 0.78rem;
+  color: var(--text);
   text-transform: uppercase;
-  letter-spacing: 0.09em;
-  margin-top: 0.25rem;
+  letter-spacing: 0.1em;
+  margin-top: 0.4rem;
+  font-weight: 500;
 }
 .stat-divider {
   width: 1px;
-  height: 34px;
-  background: var(--border);
+  height: 36px;
+  background: linear-gradient(180deg, transparent, var(--border), transparent);
   flex-shrink: 0;
 }
 
@@ -234,12 +419,36 @@ nav .logo {
   border-radius: var(--radius);
   overflow: hidden;
   text-align: left;
-  animation: fade-up 0.7s var(--ease) 0.44s both;
+  opacity: 0;
+  animation: hero-enter-terminal 1.4s var(--ease-out-expo) 1s forwards;
+  position: relative;
+}
+.terminal::before {
+  content: '';
+  position: absolute;
+  inset: -1px;
+  border-radius: var(--radius);
   box-shadow:
-    0 32px 80px rgba(0,0,0,0.6),
-    0 0 0 1px rgba(245,158,11,0.05),
-    0 0 60px rgba(245,158,11,0.04);
-  transition: opacity 0.35s ease;
+    0 0 60px rgba(245,158,11,0.05),
+    0 0 120px rgba(245,158,11,0.025),
+    0 32px 80px rgba(0,0,0,0.6);
+  pointer-events: none;
+  z-index: -1;
+}
+.terminal::after {
+  content: '';
+  position: absolute;
+  inset: 0;
+  background: repeating-linear-gradient(
+    0deg,
+    transparent,
+    transparent 2px,
+    rgba(0,0,0,0.04) 2px,
+    rgba(0,0,0,0.04) 4px
+  );
+  pointer-events: none;
+  border-radius: var(--radius);
+  z-index: 2;
 }
 .terminal-bar {
   background: var(--surface);
@@ -248,6 +457,8 @@ nav .logo {
   align-items: center;
   gap: 0.5rem;
   border-bottom: 1px solid var(--border);
+  position: relative;
+  z-index: 1;
 }
 .dot { width: 12px; height: 12px; border-radius: 50%; flex-shrink: 0; }
 .dot.red    { background: #ff5f57; }
@@ -265,6 +476,9 @@ nav .logo {
   font-size: 0.82rem;
   line-height: 1.8;
   min-height: 190px;
+  position: relative;
+  z-index: 1;
+  transition: opacity 0.35s ease;
 }
 .line-cmd   { color: var(--amber); }
 .line-out   { color: #6b7280; }
@@ -292,61 +506,142 @@ nav .logo {
 @keyframes typing { from { width: 0; } to { width: 100%; } }
 
 .hero-hint {
-  margin-top: 1.75rem;
-  font-size: 0.78rem;
-  color: var(--text-muted);
-  animation: fade-hint 3s var(--ease) both;
+  margin-top: 2rem;
+  font-size: 0.82rem;
+  color: var(--text);
+  opacity: 0;
+  animation: hero-enter 1s var(--ease-out-expo) 3s forwards;
 }
-@keyframes fade-hint {
-  0%, 60% { opacity: 0; transform: translateY(6px); }
-  100%     { opacity: 1; transform: translateY(0); }
+.hero-hint::before {
+  content: '';
+  display: block;
+  width: 1px;
+  height: 24px;
+  background: linear-gradient(180deg, transparent, var(--text-muted));
+  margin: 0 auto 0.6rem;
+  animation: hint-line 2s var(--ease) 3.2s both;
+}
+@keyframes hint-line {
+  from { height: 0; opacity: 0; }
+  to   { height: 24px; opacity: 1; }
 }
 
-/* ===================== SHARED ANIMATIONS ===================== */
-@keyframes fade-in { from { opacity: 0; } to { opacity: 1; } }
-@keyframes fade-up {
-  from { opacity: 0; transform: translateY(22px); }
+/* ===================== HERO ANIMATIONS ===================== */
+@keyframes hero-enter {
+  from { opacity: 0; transform: translateY(24px); }
   to   { opacity: 1; transform: translateY(0); }
 }
+@keyframes hero-enter-blur {
+  from { opacity: 0; transform: translateY(30px); filter: blur(8px); }
+  to   { opacity: 1; transform: translateY(0); filter: blur(0px); }
+}
+@keyframes hero-enter-terminal {
+  from { opacity: 0; transform: translateY(40px) scale(0.97); filter: blur(4px); }
+  to   { opacity: 1; transform: translateY(0) scale(1); filter: blur(0); }
+}
 
-/* Scroll reveal */
+/* ===================== SCROLL REVEAL ANIMATIONS ===================== */
+/* Base state: hidden — revealed once on scroll, stays visible */
 [data-animate] {
   opacity: 0;
-  transform: translateY(28px);
-  transition: opacity 0.6s var(--ease), transform 0.6s var(--ease);
+  transform: translateY(30px);
+  transition:
+    opacity 0.7s var(--ease),
+    transform 0.7s var(--ease);
 }
 [data-animate].visible {
   opacity: 1;
   transform: translateY(0);
 }
 
+/* Stagger children: command cards, agent cards, quotes */
+[data-animate-stagger] > * {
+  opacity: 0;
+  transform: translateY(20px);
+  transition:
+    opacity 0.5s var(--ease),
+    transform 0.5s var(--ease);
+}
+[data-animate-stagger].visible > * {
+  opacity: 1;
+  transform: translateY(0);
+}
+
+/* Stagger delay for children */
+[data-animate-stagger] > *:nth-child(1) { transition-delay: 0s; }
+[data-animate-stagger] > *:nth-child(2) { transition-delay: 0.06s; }
+[data-animate-stagger] > *:nth-child(3) { transition-delay: 0.12s; }
+[data-animate-stagger] > *:nth-child(4) { transition-delay: 0.18s; }
+[data-animate-stagger] > *:nth-child(5) { transition-delay: 0.24s; }
+[data-animate-stagger] > *:nth-child(6) { transition-delay: 0.30s; }
+
+/* Section headers — slide from left */
+[data-animate-slide] {
+  opacity: 0;
+  transform: translateX(-30px);
+  transition:
+    opacity 0.7s var(--ease),
+    transform 0.7s var(--ease);
+}
+[data-animate-slide].visible {
+  opacity: 1;
+  transform: translateX(0);
+}
+
+/* Scale-in for special blocks */
+[data-animate-scale] {
+  opacity: 0;
+  transform: scale(0.95);
+  transition:
+    opacity 0.8s var(--ease),
+    transform 0.8s var(--ease);
+}
+[data-animate-scale].visible {
+  opacity: 1;
+  transform: scale(1);
+}
+
 /* ===================== VALUE PROPS ===================== */
 .quotes {
   background: var(--surface);
   border-top: 1px solid var(--border);
   border-bottom: 1px solid var(--border);
-  padding: 4rem 2rem;
+  padding: 3.5rem 2rem;
+  position: relative;
+  overflow: hidden;
 }
 .quotes-inner {
   max-width: 980px;
   margin: 0 auto;
   display: grid;
-  grid-template-columns: repeat(auto-fit, minmax(260px, 1fr));
+  grid-template-columns: repeat(auto-fit, minmax(290px, 1fr));
   gap: 1.25rem;
 }
 .quote {
-  padding: 1.75rem;
+  padding: 2rem;
   border: 1px solid var(--border);
   border-radius: var(--radius);
   background: var(--bg);
-  transition: border-color 0.2s ease, box-shadow 0.2s ease, transform 0.2s ease;
+  position: relative;
+  overflow: hidden;
+  transition: border-color 0.3s ease, transform 0.3s ease, box-shadow 0.3s ease;
+}
+.quote .card-glow {
+  position: absolute;
+  width: 300px; height: 300px;
+  background: radial-gradient(circle, rgba(245,158,11,0.07) 0%, transparent 70%);
+  pointer-events: none;
+  transform: translate(-50%, -50%);
+  opacity: 0;
+  transition: opacity 0.3s ease;
 }
 .quote:hover {
-  border-color: rgba(245,158,11,0.3);
-  box-shadow: 0 0 28px rgba(245,158,11,0.07);
+  border-color: rgba(245,158,11,0.25);
   transform: translateY(-3px);
+  box-shadow: 0 8px 32px rgba(245,158,11,0.04);
 }
-.quote p { color: var(--text-muted); font-size: 0.875rem; line-height: 1.75; }
+.quote:hover .card-glow { opacity: 1; }
+.quote p { color: var(--text); font-size: 0.92rem; line-height: 1.75; }
 .quote p strong {
   color: var(--text);
   display: block;
@@ -355,27 +650,41 @@ nav .logo {
 }
 
 /* ===================== SECTIONS ===================== */
-section { padding: 5.5rem 2rem; }
-.container { max-width: 980px; margin: 0 auto; }
+section {
+  padding: 4rem 2rem;
+  position: relative;
+  overflow: hidden;
+}
+.container { max-width: 980px; margin: 0 auto; position: relative; z-index: 2; }
 .section-label {
   font-family: var(--font-mono);
-  font-size: 0.68rem;
+  font-size: 0.76rem;
   color: var(--amber);
   text-transform: uppercase;
   letter-spacing: 0.14em;
   margin-bottom: 0.75rem;
+  display: flex;
+  align-items: center;
+  gap: 0.75rem;
+}
+.section-label::after {
+  content: '';
+  flex: 1;
+  height: 1px;
+  background: linear-gradient(90deg, rgba(245,158,11,0.3), transparent 60%);
+  max-width: 120px;
 }
 h2 {
-  font-size: clamp(1.5rem, 3vw, 2.15rem);
-  font-weight: 700;
+  font-family: var(--font-sans);
+  font-size: clamp(1.8rem, 3.5vw, 2.5rem);
+  font-weight: 600;
   margin-bottom: 1rem;
-  letter-spacing: -0.025em;
+  letter-spacing: -0.02em;
 }
 .section-sub {
-  color: var(--text-muted);
-  max-width: 520px;
-  margin-bottom: 3rem;
-  font-size: 0.95rem;
+  color: var(--text);
+  margin-bottom: 2.5rem;
+  font-size: 1rem;
   line-height: 1.65;
 }
 .inline-code {
@@ -385,41 +694,66 @@ h2 {
 }
 
 /* ===================== HOW IT WORKS ===================== */
+#how-it-works {
+  position: relative;
+}
 .steps {
   display: grid;
-  grid-template-columns: repeat(auto-fit, minmax(240px, 1fr));
+  grid-template-columns: repeat(auto-fit, minmax(270px, 1fr));
   gap: 1.25rem;
 }
 .step {
-  padding: 1.75rem;
+  padding: 2rem 1.75rem;
   border: 1px solid var(--border);
   border-radius: var(--radius);
   background: var(--bg);
   position: relative;
-  transition: border-color 0.2s ease, box-shadow 0.25s ease, transform 0.2s ease;
+  overflow: hidden;
+  transition: border-color 0.3s ease, transform 0.3s ease, box-shadow 0.3s ease;
 }
-.step::before {
-  content: '';
+.step .card-glow {
   position: absolute;
-  inset: 0;
-  border-radius: var(--radius);
-  background: linear-gradient(135deg, var(--amber-glow), transparent);
+  width: 300px; height: 300px;
+  background: radial-gradient(circle, rgba(245,158,11,0.07) 0%, transparent 70%);
+  pointer-events: none;
+  transform: translate(-50%, -50%);
   opacity: 0;
-  transition: opacity 0.25s ease;
+  transition: opacity 0.3s ease;
 }
-.step:hover { border-color: rgba(245,158,11,0.35); transform: translateY(-3px); }
-.step:hover::before { opacity: 1; }
+.step:hover {
+  border-color: rgba(245,158,11,0.3);
+  transform: translateY(-3px);
+  box-shadow: 0 8px 32px rgba(245,158,11,0.04);
+}
+.step:hover .card-glow { opacity: 1; }
 .step-num {
   font-family: var(--font-mono);
-  font-size: 2.2rem;
+  font-size: 2.8rem;
   font-weight: 700;
   color: var(--amber);
-  opacity: 0.22;
+  opacity: 0.5;
   margin-bottom: 0.75rem;
   line-height: 1;
+  position: relative;
+  text-shadow: 0 0 24px rgba(245,158,11,0.3);
+}
+.step h3 { font-family: var(--font-sans); font-size: 1.1rem; font-weight: 600; margin-bottom: 0.5rem; position: relative; }
+.step p  { font-size: 0.92rem; color: var(--text); position: relative; }
+
+/* Connector line between steps on large screens */
+@media (min-width: 768px) {
+  .steps { position: relative; }
+  .step:not(:last-child)::after {
+    content: '';
+    position: absolute;
+    right: -0.75rem;
+    top: 50%;
+    width: 1.25rem;
+    height: 1px;
+    background: linear-gradient(90deg, rgba(245,158,11,0.25), transparent);
+    pointer-events: none;
+  }
 }
-.step h3 { font-size: 1rem; margin-bottom: 0.5rem; position: relative; }
-.step p  { font-size: 0.875rem; color: var(--text-muted); position: relative; }
 
 /* ===================== COMMANDS ===================== */
 .commands-section {
@@ -427,10 +761,10 @@ h2 {
   border-top: 1px solid var(--border);
   border-bottom: 1px solid var(--border);
 }
-.phase-group { margin-bottom: 2.75rem; }
+.phase-group { margin-bottom: 3rem; }
 .phase-label {
   font-family: var(--font-mono);
-  font-size: 0.67rem;
+  font-size: 0.76rem;
   color: var(--amber);
   text-transform: uppercase;
   letter-spacing: 0.13em;
@@ -440,7 +774,7 @@ h2 {
 }
 .command-grid {
   display: grid;
-  grid-template-columns: repeat(auto-fill, minmax(260px, 1fr));
+  grid-template-columns: repeat(auto-fill, minmax(290px, 1fr));
   gap: 0.875rem;
 }
 .command-card {
@@ -449,96 +783,143 @@ h2 {
   border-radius: var(--radius);
   background: var(--bg);
   cursor: default;
-  transition: border-color 0.2s ease, box-shadow 0.25s ease, transform 0.2s ease;
+  position: relative;
+  overflow: hidden;
+  transition: border-color 0.3s ease, transform 0.25s ease, box-shadow 0.3s ease;
+}
+.command-card .card-glow {
+  position: absolute;
+  width: 250px; height: 250px;
+  background: radial-gradient(circle, rgba(245,158,11,0.06) 0%, transparent 70%);
+  pointer-events: none;
+  transform: translate(-50%, -50%);
+  opacity: 0;
+  transition: opacity 0.3s ease;
 }
 .command-card:hover {
-  border-color: rgba(245,158,11,0.45);
-  box-shadow: 0 0 22px rgba(245,158,11,0.09), inset 0 0 22px rgba(245,158,11,0.02);
+  border-color: rgba(245,158,11,0.35);
   transform: translateY(-2px);
+  box-shadow: 0 4px 20px rgba(245,158,11,0.03);
 }
+.command-card:hover .card-glow { opacity: 1; }
 .command-card code {
   font-family: var(--font-mono);
-  font-size: 0.74rem;
+  font-size: 0.82rem;
   color: var(--amber);
   display: block;
   margin-bottom: 0.5rem;
+  position: relative;
 }
 .command-card p {
-  font-size: 0.82rem;
-  color: var(--text-muted);
+  font-size: 0.88rem;
+  color: var(--text);
   line-height: 1.55;
+  position: relative;
 }
 
 /* ===================== AGENTS ===================== */
 .agent-grid {
   display: grid;
-  grid-template-columns: repeat(auto-fit, minmax(260px, 1fr));
+  grid-template-columns: repeat(auto-fit, minmax(290px, 1fr));
   gap: 1.25rem;
 }
 .agent-card {
-  padding: 1.75rem;
+  padding: 2rem 1.75rem;
   border: 1px solid var(--border);
   border-radius: var(--radius);
   background: var(--bg);
-  transition: border-color 0.2s ease, box-shadow 0.25s ease, transform 0.2s ease;
+  position: relative;
+  overflow: hidden;
+  transition: border-color 0.3s ease, transform 0.3s ease, box-shadow 0.3s ease;
+}
+.agent-card .card-glow {
+  position: absolute;
+  width: 300px; height: 300px;
+  background: radial-gradient(circle, rgba(245,158,11,0.07) 0%, transparent 70%);
+  pointer-events: none;
+  transform: translate(-50%, -50%);
+  opacity: 0;
+  transition: opacity 0.3s ease;
 }
 .agent-card:hover {
-  border-color: rgba(245,158,11,0.4);
-  box-shadow: 0 0 36px rgba(245,158,11,0.09);
+  border-color: rgba(245,158,11,0.35);
   transform: translateY(-3px);
+  box-shadow: 0 8px 32px rgba(245,158,11,0.04);
 }
+.agent-card:hover .card-glow { opacity: 1; }
 .agent-icon {
   font-size: 1.4rem;
   color: var(--amber);
-  opacity: 0.55;
+  opacity: 0.45;
   margin-bottom: 0.75rem;
   display: block;
+  transition: opacity 0.3s ease, transform 0.3s ease;
 }
+.agent-card:hover .agent-icon { opacity: 0.8; transform: rotate(45deg); }
 .agent-card h3 {
   font-family: var(--font-mono);
   color: var(--amber);
   margin-bottom: 0.4rem;
   font-size: 0.9rem;
+  position: relative;
 }
 .agent-card .role {
-  font-size: 0.85rem;
+  font-family: var(--font-sans);
+  font-size: 0.92rem;
   font-weight: 600;
   color: var(--text);
   margin-bottom: 0.75rem;
+  position: relative;
 }
-.agent-card p { font-size: 0.825rem; color: var(--text-muted); line-height: 1.65; }
+.agent-card p { font-size: 0.9rem; color: var(--text); line-height: 1.65; position: relative; }
 
 /* v2 callout */
 .v2-callout {
-  background: linear-gradient(135deg, rgba(245,158,11,0.07), rgba(245,158,11,0.01) 60%);
-  border: 1px solid rgba(245,158,11,0.22);
+  background: linear-gradient(135deg, rgba(245,158,11,0.06), rgba(245,158,11,0.01) 60%);
+  border: 1px solid rgba(245,158,11,0.18);
   border-radius: 12px;
   padding: 2.5rem;
-  margin-top: 2rem;
+  margin-top: 2.5rem;
+  position: relative;
+  overflow: hidden;
+}
+.v2-callout::before {
+  content: '2.0';
+  position: absolute;
+  right: 2rem;
+  top: 1.5rem;
+  font-family: var(--font-display);
+  font-size: 5rem;
+  font-weight: 800;
+  text-transform: uppercase;
+  color: var(--amber);
+  opacity: 0.05;
+  pointer-events: none;
+  line-height: 1;
 }
 .v2-callout h3 {
   font-family: var(--font-mono);
   color: var(--amber);
   margin-bottom: 1.25rem;
-  font-size: 0.88rem;
+  font-size: 0.86rem;
   letter-spacing: 0.06em;
   text-transform: uppercase;
 }
 .v2-callout ul {
   list-style: none;
   display: grid;
-  grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
+  grid-template-columns: repeat(auto-fit, minmax(280px, 1fr));
   gap: 0.625rem;
 }
 .v2-callout li {
-  font-size: 0.875rem;
-  color: var(--text-muted);
+  font-size: 0.92rem;
+  color: var(--text);
   padding-left: 1.1rem;
   position: relative;
   line-height: 1.55;
 }
 .v2-callout li::before {
-  content: '→';
+  content: '\2192';
   position: absolute;
   left: 0;
   color: var(--amber);
@@ -561,22 +942,41 @@ h2 {
   white-space: pre;
   overflow-x: auto;
   line-height: 1.9;
+  transition: border-color 0.3s ease, box-shadow 0.3s ease;
+}
+.code-block:hover {
+  border-color: rgba(245,158,11,0.2);
+  box-shadow: 0 0 40px rgba(245,158,11,0.04);
 }
 .code-block .comment { color: var(--text-muted); }
 .code-block .cmd     { color: var(--amber); }
-.install-note { font-size: 0.875rem; color: var(--text-muted); }
+.install-note { font-size: 0.92rem; color: var(--text); }
 
 /* ===================== FOOTER ===================== */
 footer {
   border-top: 1px solid var(--border);
-  padding: 2rem;
+  padding: 2.5rem;
   text-align: center;
-  color: var(--text-muted);
-  font-size: 0.8rem;
+  color: var(--text);
+  font-size: 0.84rem;
+  letter-spacing: 0.01em;
 }
 footer a { color: var(--text-muted); text-decoration: none; transition: color 0.18s ease; }
 footer a:hover { color: var(--amber); }
 
+/* ===================== SCROLL PROGRESS BAR ===================== */
+.scroll-progress {
+  position: fixed;
+  top: 0;
+  left: 0;
+  height: 2px;
+  background: linear-gradient(90deg, var(--amber-dim), var(--amber), var(--amber-dim));
+  z-index: 101;
+  width: 0%;
+  transition: width 0.05s linear;
+  box-shadow: 0 0 8px rgba(245,158,11,0.4);
+}
+
 /* ===================== RESPONSIVE ===================== */
 @media (max-width: 640px) {
   .nav-links a { display: none; }
@@ -585,6 +985,7 @@ footer a:hover { color: var(--amber); }
   .stat-num { font-size: 1.3rem; }
   .stat-divider { height: 24px; }
   .terminal-body { font-size: 0.75rem; }
+  .hero h1 { font-size: clamp(1.8rem, 8vw, 2.8rem); }
 }
 
 /* ===================== REDUCED MOTION ===================== */
@@ -593,7 +994,16 @@ footer a:hover { color: var(--amber); }
     animation-duration: 0.01ms !important;
     transition-duration: 0.01ms !important;
   }
-  [data-animate] { opacity: 1; transform: none; }
+  [data-animate],
+  [data-animate-stagger] > *,
+  [data-animate-slide],
+  [data-animate-scale] { opacity: 1; transform: none; }
   .shimmer { animation: none; color: var(--amber); -webkit-text-fill-color: var(--amber); }
-  .hero-glow { animation: none; }
+  .hero-badge, .hero h1, .hero .tagline,
+  .install-block, .stats-strip, .terminal, .hero-hint {
+    opacity: 1; transform: none; filter: none;
+  }
+  #hero-particles, #ambient-particles { display: none; }
+  .scroll-progress { display: none; }
+  .aurora, .light-sweep, .radial-pulse, .shimmer-line { display: none; }
 }

From c53d9314916a5fb65dff28c7fdde179807018a2f Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Sat, 7 Mar 2026 05:08:19 +0530
Subject: [PATCH 12/15] docs: update skill descriptions and settings

Signed-off-by: Claude Opus 4.6 <noreply@anthropic.com>
Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 .claude/settings.local.json          |  11 +-
 skills/accessibility/SKILL.md        | 161 ++++++++++++++++++++++-----
 skills/api-test/SKILL.md             |   2 +-
 skills/bug-report/SKILL.md           |   2 +-
 skills/coverage-gap/SKILL.md         |   2 +-
 skills/e2e-test/SKILL.md             |  16 +--
 skills/exploratory-testing/SKILL.md  |   2 +-
 skills/flaky-test-diagnosis/SKILL.md |   2 +-
 skills/pr-review/SKILL.md            |   2 +-
 skills/regression-planner/SKILL.md   |   2 +-
 skills/release-readiness/SKILL.md    | 140 ++++++++++++++++++-----
 skills/risk-prioritization/SKILL.md  |   2 +-
 skills/setup/SKILL.md                |  36 ++++--
 skills/test-cases/SKILL.md           |   2 +-
 skills/test-data/SKILL.md            |   2 +-
 skills/test-plan/SKILL.md            |   2 +-
 16 files changed, 302 insertions(+), 84 deletions(-)

diff --git a/.claude/settings.local.json b/.claude/settings.local.json
index e945dca..02f631d 100644
--- a/.claude/settings.local.json
+++ b/.claude/settings.local.json
@@ -1,10 +1,17 @@
 {
   "permissions": {
     "allow": [
-      "Bash(git:*)",
+      "Bash(git diff:*)",
+      "Bash(git log:*)",
+      "Bash(git show:*)",
+      "Bash(git branch:*)",
+      "Bash(git status:*)",
+      "Bash(git fetch:*)",
       "Bash(node scripts/detect-project.js:*)",
       "Bash(node scripts/session-hook.js:*)",
-      "Bash(node scripts/state-manager.js:*)"
+      "Bash(node scripts/state-manager.js:*)",
+      "Bash(npm install:*)",
+      "Bash(node:*)"
     ]
   }
 }
diff --git a/skills/accessibility/SKILL.md b/skills/accessibility/SKILL.md
index 2a9a1e0..54761a0 100644
--- a/skills/accessibility/SKILL.md
+++ b/skills/accessibility/SKILL.md
@@ -1,40 +1,151 @@
 ---
 name: accessibility
-description: Guided WCAG 2.1 accessibility audit with plain-English manual test scripts
+description: WCAG 2.1 accessibility audit with plain-English manual test scripts and stack-matched automated tool recommendations. Use this skill whenever a user wants to audit accessibility, asks "is this accessible?", needs an a11y audit of a page or component, or wants to check WCAG compliance before a release.
 ---
 
 # Accessibility Checker
 
-Guided accessibility audit against WCAG 2.1. Read `qa-artifacts/.qa-config.json` for project context.
+Conduct a WCAG 2.1 Level AA accessibility audit. Reads project and session state to tailor automated tool recommendations to the detected stack and surface prior audit history. Generates manual test scripts anyone can follow, categorized findings, and actionable recommendations.
 
 ## Input
-Accept via `$ARGUMENTS`: page URL, feature name, or component description.
+Accept via `$ARGUMENTS`: page URL, feature name, component description, or user flow.
 
 ## Workflow
 
-1. **Select audit scope**: full page, specific component, or entire user flow
-2. **Run guided audit** across WCAG categories:
-   - **Perceivable**: images have alt text, color contrast, text alternatives
-   - **Operable**: keyboard navigation, focus management, timing
-   - **Understandable**: labels, error messages, predictable behavior
-   - **Robust**: semantic HTML, ARIA usage, screen reader compatibility
-3. **Generate manual test scripts** in plain English
-4. **Recommend automated tools** for the detected tech stack
-5. **Produce audit report**
-
-## Manual Test Scripts
-Generate step-by-step instructions anyone can follow:
+### Step 1: Read State
+
+Execute state reading commands:
+
+```bash
+PROJECT_STATE=$(node scripts/state-manager.js read project)
+SESSION_STATE=$(node scripts/state-manager.js read session)
+DETECTION=$(node scripts/state-manager.js read project detection)
+SKILL_HISTORY=$(node scripts/state-manager.js read session skillHistory)
+FEATURE=$(node scripts/state-manager.js read session featureUnderTest)
+```
+
+Extract from project state:
+- `detection.languages`, `detection.frameworks` — for automated tool recommendations (axe-playwright for Playwright, jest-axe for Jest, etc.)
+- `detection.testFrameworks` — for recommending runnable accessibility assertions
+
+Extract from session state:
+- `featureUnderTest` — for scope context if not specified in `$ARGUMENTS`
+- `skillHistory[]` — check for prior accessibility runs on the same area
+
+If no state available, proceed with general tool recommendations and standard WCAG audit.
+
+### Step 2: Parse Input and Determine Scope
+
+From `$ARGUMENTS`, identify:
+- **Audit target** — page, component, feature area, or user flow
+- **Scope** — full page, specific component, or entire user flow
+
+Derive save label (normalize: lowercase, hyphens only).
+
+### Step 3: Check Prior Output
+
+Check session `skillHistory` for prior `accessibility` invocations:
+- If prior found for same target: add note: "Note: An accessibility audit for {target} was already run this session. Generating fresh audit — compare findings for resolution tracking."
+
+### Step 4: Run Guided Audit
+
+Audit across all four WCAG POUR principles. For each finding, record: WCAG criterion, severity, location, and recommendation.
+
+**Perceivable**
+- Images: alt text present and descriptive (not filename-only or empty for non-decorative images)
+- Color: text meets 4.5:1 contrast ratio (3:1 for large text)
+- Text alternatives: non-text content has text equivalents
+- Audio/video: captions and transcripts where applicable
+
+**Operable**
+- Keyboard: all functionality reachable and operable without a mouse
+- Focus: visible focus indicator on all interactive elements; no keyboard traps
+- Skip links: skip navigation present for repetitive content
+- Timing: no time limits that cannot be extended; animations can be paused
+
+**Understandable**
+- Labels: all form fields have associated labels
+- Error messages: validation errors identify the field and explain how to fix
+- Instructions: complex interactions have clear instructions
+- Predictable: pages behave consistently; no unexpected context changes on focus
+
+**Robust**
+- Semantic HTML: headings, landmarks, lists used correctly
+- ARIA: roles/states/properties used correctly; no redundant or conflicting ARIA
+- Screen reader: content announced in logical order; dynamic updates announced
+
+For full WCAG criterion details and testing notes, read `references/wcag-checklist.md`.
+
+### Step 5: Generate Manual Test Scripts
+
+Produce numbered step-by-step instructions a non-developer can follow. For each test, include a clear pass/fail condition.
+
 ```
 1. TAB through the page from top to bottom
    ✅ PASS if: every interactive element receives visible focus
    ❌ FAIL if: focus disappears, skips elements, or has no visible indicator
 
 2. With a screen reader (NVDA/VoiceOver), navigate the page
-   ✅ PASS if: all content is announced in logical order
-   ❌ FAIL if: images are skipped, buttons are unlabeled, or content is repeated
+   ✅ PASS if: all content announced in logical order, buttons have descriptive labels
+   ❌ FAIL if: images skipped, buttons unlabeled, or content repeated
+```
+
+For detailed manual test procedures per WCAG criterion, read `references/manual-tests.md`.
+
+### Step 6: Recommend Automated Tools
+
+Based on detected stack:
+- **Playwright detected**: `@axe-core/playwright` or `axe-playwright` — runnable in existing test suite
+- **Jest detected**: `jest-axe` — axe-core assertions in unit/component tests
+- **React/Next.js detected**: `@axe-core/react` for development-time feedback
+- **No framework detected**: Browser extension `axe DevTools`, `WAVE`, or Lighthouse (built into Chrome DevTools)
+
+### Step 7: Format with Context Preamble
+
+Generate context preamble:
+
+```markdown
+> **Project Context**
+> Stack: {languages} | Frameworks: {frameworks} | Test Framework: {testFrameworks}
+>
+> **Why this audit:** {framework}-specific tool recommendations included.
+> Prior audits this session: {N or 'none'}.
 ```
 
+Cold start: `Stack: Not detected — general tool recommendations provided.`
+
+### Step 8: Write Session State
+
+```bash
+node scripts/state-manager.js write session skillHistory '{
+  "skill": "accessibility",
+  "feature": "<audit-target>",
+  "timestamp": "<ISO-8601>",
+  "auditTarget": "<page-or-component>",
+  "criticalCount": <n>,
+  "majorCount": <n>,
+  "minorCount": <n>,
+  "wcagLevel": "AA"
+}'
+
+node scripts/state-manager.js merge session findings '{
+  "area": "<audit-target>",
+  "type": "accessibility",
+  "summary": "A11y audit: <critical> critical, <major> major, <minor> minor issues against WCAG 2.1 AA"
+}'
+```
+
+## Output Section Order
+
+1. **Context preamble** — blockquote with project context
+2. **Prior output note** — if applicable (Step 3)
+3. **Audit findings table** — categorized by POUR principle with severity
+4. **Manual Test Scripts** — numbered plain-English steps with pass/fail criteria
+5. **Recommended Automated Tools** — stack-matched
+6. **Suggested Next Steps**
+
 ## Output Structure
+
 ```
 ## Accessibility Audit: <page/feature>
 Date: YYYY-MM-DD
@@ -50,18 +161,16 @@ Standard: WCAG 2.1 Level AA
 |---|-------|---------------|----------|----------|----------------|
 
 ### Manual Test Scripts
-<numbered plain-language test steps>
+<numbered plain-language test steps with pass/fail criteria>
 
-### Recommended Tools
-<based on project tech stack>
+### Recommended Automated Tools
+<based on detected stack>
 ```
 
-For full WCAG checklist, read `references/wcag-checklist.md`.
-For manual test procedures, read `references/manual-tests.md`.
-
 ## Save
-Save to `qa-artifacts/a11y-audits/a11y-YYYY-MM-DD-<page>.md`
+Save to `qa-artifacts/a11y-audits/a11y-YYYY-MM-DD-<target>.md`
 
 ## Suggested Next Steps
-After generating the audit, suggest:
-- "File each critical/major finding as a bug report with `/qa-toolkit:bug-report`."
+After generating the audit:
+- "File each Critical/Major finding as a bug report with `/qa-toolkit:bug-report`."
+- If E2E framework detected: "Add automated axe assertions to your Playwright tests with `/qa-toolkit:e2e-test`."
diff --git a/skills/api-test/SKILL.md b/skills/api-test/SKILL.md
index 7a8aa65..dd3e2bd 100644
--- a/skills/api-test/SKILL.md
+++ b/skills/api-test/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: api-test
-description: Generate state-aware API test suites with contract validation from OpenAPI/Swagger specs as cURL, Postman, or Playwright.
+description: Generate API test suites from endpoint descriptions, OpenAPI/Swagger specs, or workflow descriptions — as cURL commands, Postman collections, or Playwright API tests with contract validation. Use this skill whenever a user wants to test an API endpoint, verify REST API behavior, validate against a spec, or asks for API test coverage.
 ---
 
 # API Test Builder
diff --git a/skills/bug-report/SKILL.md b/skills/bug-report/SKILL.md
index cef8f8b..5b953c9 100644
--- a/skills/bug-report/SKILL.md
+++ b/skills/bug-report/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: bug-report
-description: Generate structured bug reports with environment auto-fill from project state, severity boost from risk areas, and session state write-back.
+description: Transform any bug observation into a structured, tracker-ready bug report with auto-filled environment, severity classification, and duplicate search terms. Use this skill whenever a user describes something broken, unexpected behavior, a crash, an error message, or says things like "it doesn't work", "I found a bug", or "something's wrong with X" — even a single sentence produces a complete, ready-to-file report.
 ---
 
 # Bug Report Generator
diff --git a/skills/coverage-gap/SKILL.md b/skills/coverage-gap/SKILL.md
index 7e5e2f9..0674ef7 100644
--- a/skills/coverage-gap/SKILL.md
+++ b/skills/coverage-gap/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: coverage-gap
-description: Analyze test coverage against requirements and code paths to identify untested areas with severity classification.
+description: Identify untested areas by comparing test files against source modules, tracing requirements to tests, and detecting missing test types — with severity classification per gap. Use this skill when a user asks "what aren't we testing?", wants gap analysis before a release, or needs to map requirements to test coverage.
 ---
 
 # Coverage Gap Analyzer
diff --git a/skills/e2e-test/SKILL.md b/skills/e2e-test/SKILL.md
index 76f466d..8d018da 100644
--- a/skills/e2e-test/SKILL.md
+++ b/skills/e2e-test/SKILL.md
@@ -18,7 +18,7 @@ Accept via `$ARGUMENTS`: scenario descriptions and optional flags. Examples:
 
 ### Step 1: Read State
 
-Read `skills/shared/references/state-integration.md` and execute the state reading commands:
+Read `shared/references/state-integration.md` and execute the state reading commands:
 
 ```bash
 PROJECT_STATE=$(node scripts/state-manager.js read project)
@@ -31,7 +31,7 @@ FEATURE=$(node scripts/state-manager.js read session featureUnderTest)
 
 Extract: tech stack, test frameworks, test directories, conventions, session feature, prior skill history.
 
-If no state available, use defaults per `skills/shared/references/cold-start-pattern.md`:
+If no state available, use defaults per `shared/references/cold-start-pattern.md`:
 - languages: [] (generic examples)
 - testFrameworks: [] (Playwright best-practice defaults)
 - featureUnderTest: null (derive from $ARGUMENTS)
@@ -117,7 +117,7 @@ For advanced patterns (fixture integration, POM best practices, waiting strategi
 
 ### Step 8: Generate Context Preamble
 
-Read `skills/shared/references/context-preamble.md` and generate the preamble blockquote.
+Read `shared/references/context-preamble.md` and generate the preamble blockquote.
 
 Include convention scan results in the preamble:
 - Which files were scanned for conventions
@@ -160,7 +160,7 @@ node scripts/state-manager.js merge session findings '{
 }'
 ```
 
-Check for prior invocations per `skills/shared/references/state-integration.md` section 4 (Prior Output Awareness). If same feature was tested before, note it but generate full output.
+Check for prior invocations per `shared/references/state-integration.md` section 4 (Prior Output Awareness). If same feature was tested before, note it but generate full output.
 
 ## Output Sections
 
@@ -173,7 +173,7 @@ Produce output in this order:
 5. **New page object scaffolds** -- if any, in separate code blocks matching existing POM style
 6. **Fixture updates** -- if custom fixtures needed, show additions to existing `fixtures.ts`
 7. **Run instructions** -- reference `scripts/run-test.js` with `--help` for options
-8. **Cold-start footer** -- if applicable, per `skills/shared/references/cold-start-pattern.md`
+8. **Cold-start footer** -- if applicable, per `shared/references/cold-start-pattern.md`
 9. **Suggested Next Steps** -- contextual recommendations
 
 ## Prior Output Awareness
@@ -189,11 +189,11 @@ Always generate full output regardless.
 
 | Reference | When to Load |
 |---|---|
-| `skills/shared/references/state-integration.md` | Always (step 1) |
-| `skills/shared/references/context-preamble.md` | Always (step 8) |
+| `shared/references/state-integration.md` | Always (step 1) |
+| `shared/references/context-preamble.md` | Always (step 8) |
 | `references/convention-scanner.md` | When Playwright detected AND no cached conventions (step 3) |
 | `references/playwright-patterns.md` | Always for advanced patterns (step 7) |
-| `skills/shared/references/cold-start-pattern.md` | When state reading returns empty (step 1) |
+| `shared/references/cold-start-pattern.md` | When state reading returns empty (step 1) |
 
 ## Save
 
diff --git a/skills/exploratory-testing/SKILL.md b/skills/exploratory-testing/SKILL.md
index 1a1440d..304ad41 100644
--- a/skills/exploratory-testing/SKILL.md
+++ b/skills/exploratory-testing/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: exploratory-testing
-description: Guide structured exploratory testing sessions with charters, heuristic prompts, time-boxing guidance, and session notes capture.
+description: Generate a complete exploratory testing session package — charter, filtered heuristic prompts, time-boxing check-ins, and session notes templates — in one shot. Use this skill when a user wants to do unscripted testing, asks "how do I explore this feature manually?", or needs a structured charter for session-based testing of any feature or user flow.
 ---
 
 # Exploratory Testing Session Guide
diff --git a/skills/flaky-test-diagnosis/SKILL.md b/skills/flaky-test-diagnosis/SKILL.md
index efdeda3..091625f 100644
--- a/skills/flaky-test-diagnosis/SKILL.md
+++ b/skills/flaky-test-diagnosis/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: flaky-test-diagnosis
-description: Diagnose flaky tests by analyzing test code and failure logs to identify root causes with specific fix recommendations.
+description: Diagnose flaky tests with root cause classification (timing, shared state, environment, etc.) and concrete fix recommendations with code examples. Use this skill whenever a user mentions intermittently failing tests, CI flakiness, tests that "sometimes pass", or shares test output with inconsistent results — works with test code alone or with failure logs.
 ---
 
 # Flaky Test Diagnoser
diff --git a/skills/pr-review/SKILL.md b/skills/pr-review/SKILL.md
index 51436ca..5bf0318 100644
--- a/skills/pr-review/SKILL.md
+++ b/skills/pr-review/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: pr-review
-description: Analyze PR changes with state-aware risk assessment, change-impact analysis, and QA checklist grounded in project context.
+description: Analyze PR changes with risk assessment, change-impact analysis, and a diff-specific QA checklist grounded in project context. Use this skill whenever a user shares a PR link, branch name, diff, or list of changed files and wants to know what to test — even a casual description like "I just changed the auth module" is enough to start a review.
 ---
 
 # PR Review Assistant
diff --git a/skills/regression-planner/SKILL.md b/skills/regression-planner/SKILL.md
index d38b7c3..31c247f 100644
--- a/skills/regression-planner/SKILL.md
+++ b/skills/regression-planner/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: regression-planner
-description: Analyze change impact with state-aware risk ranking, dependency awareness, and cross-skill consumption of prior PR review and bug report findings.
+description: Plan targeted regression testing by mapping impact radius, ranking risk, and recommending scope (Quick/Standard/Full). Use this skill when a user asks "what should we retest?", has just made changes and needs to know what's at risk, or is preparing for a release and needs to scope regression effort from a PR, file list, or feature description.
 ---
 
 # Regression Test Planner
diff --git a/skills/release-readiness/SKILL.md b/skills/release-readiness/SKILL.md
index 5949add..dfa96c0 100644
--- a/skills/release-readiness/SKILL.md
+++ b/skills/release-readiness/SKILL.md
@@ -1,43 +1,122 @@
 ---
 name: release-readiness
-description: Go/no-go release assessment with quality gate scoring and risk analysis
+description: Go/no-go release assessment with interactive quality gate scoring, risk matrix, and monitoring plan. Use whenever a team is deciding whether to release, asks "are we ready to ship?", or needs a formal sign-off document before any release from hotfix to major version.
 ---
 
 # Release Readiness Assessment
 
-Generate a go/no-go release decision. Read `qa-artifacts/.qa-config.json` for project context.
+Generate a go/no-go release decision through an interactive quality gate assessment. Reads project and session state to pre-populate quality signals from prior skill findings (regression plan, PR reviews, bug reports) — reducing questions asked. Produces a scored verdict with conditions, risk matrix, monitoring plan, and rollback criteria.
 
 ## Input
-Accept via `$ARGUMENTS`: release type, version, or just start an interactive assessment.
+Accept via `$ARGUMENTS`: release type (hotfix/patch/minor/major), version number, or start without arguments.
 
 ## Workflow
 
-1. **Identify release type**:
-   - Hotfix (emergency fix, minimal testing)
-   - Patch (bug fixes, targeted testing)
-   - Minor (new features, standard testing)
-   - Major (breaking changes, full regression)
-2. **Gather quality signals** — ask the user about each:
-   - Test execution status — how many tests passed/failed/blocked?
-   - Known open bugs — any blockers or criticals?
-   - New features tested? — against acceptance criteria?
-   - Regression testing — completed to what scope?
-   - Performance — any degradation noted?
-   - Security — any new vulnerabilities?
-3. **Score each quality gate** (0-100):
-   - Test Coverage Score
-   - Bug Status Score (based on open severity)
-   - Regression Score
-   - Deployment Readiness Score
-4. **Generate recommendation**
+### Step 1: Read State
+
+```bash
+PROJECT_STATE=$(node scripts/state-manager.js read project)
+SESSION_STATE=$(node scripts/state-manager.js read session)
+DETECTION=$(node scripts/state-manager.js read project detection)
+RISKS=$(node scripts/state-manager.js read project risks)
+SKILL_HISTORY=$(node scripts/state-manager.js read session skillHistory)
+FINDINGS=$(node scripts/state-manager.js read session findings)
+```
+
+Extract from project state:
+- `risks[]` — pre-populate risk matrix entries
+- `detection.testFrameworks` — for deployment readiness checks
+
+Extract from session state:
+- `skillHistory[]` — check for `regression-planner`, `pr-review`, `bug-report` entries to pre-fill quality signals
+- `findings[]` — risk or bug findings inform quality gate scoring
+
+### Step 2: Identify Release Type
+
+From `$ARGUMENTS` or by asking the user:
+- **Hotfix** — emergency fix, minimal testing required
+- **Patch** — bug fixes, targeted testing
+- **Minor** — new features, standard testing
+- **Major** — breaking changes, full regression
+
+Read `references/release-types.md` for type-specific quality gate thresholds.
+
+### Step 3: Pre-fill Quality Signals from Session
+
+Before asking questions, extract available signals:
+- **If `regression-planner` in skillHistory**: use its `recommendedScope` and `p0Areas` to pre-answer the regression testing question
+- **If `pr-review` in skillHistory**: use its `riskLevel` and `regressionSurface` for test coverage and risk context
+- **If `bug-report` in skillHistory**: use `severity` entries to pre-populate known open bugs
+- **If `risks[]` in project state**: surface in risk matrix as pre-populated entries
+
+For each pre-filled signal, state it explicitly and ask the user to confirm or correct rather than asking from scratch.
+
+### Step 4: Gather Remaining Quality Signals
+
+For signals not available from session state, ask the user:
+- Test execution status — how many tests passed/failed/blocked?
+- Known open bugs — any blockers or criticals not yet in session?
+- New features tested against acceptance criteria?
+- Performance — any degradation noted?
+- Security — any new vulnerabilities?
+
+### Step 5: Score Quality Gates (0–100)
+
+| Gate | What it measures |
+|------|-----------------|
+| Test Coverage | % of P0 tests passing, coverage of changed areas |
+| Bug Status | Open blocker/critical bugs (blocker = 0 score, each critical = -10) |
+| Regression | Regression scope completed vs recommended scope |
+| Deployment Readiness | Env configs, feature flags, rollback plan in place |
+
+For thresholds per release type, read `references/release-types.md`.
+
+### Step 6: Generate Verdict
+
+- **GO**: All gates ≥ 70, no open blockers
+- **CONDITIONAL GO**: One or more gates 50–69, or open criticals with mitigation plan
+- **NO-GO**: Any gate < 50, or any open blocker
+
+### Step 7: Format with Context Preamble
+
+```markdown
+> **Project Context**
+> Stack: {languages} | Frameworks: {frameworks}
+> Prior session: {N findings from regression-planner/pr-review/bug-report used to pre-fill}
+>
+> **Why this assessment:** {N} prior skill findings used to pre-populate quality signals.
+```
+
+Cold start: "No prior session context — all quality signals gathered interactively."
+
+### Step 8: Write Session State
+
+```bash
+node scripts/state-manager.js write session skillHistory '{
+  "skill": "release-readiness",
+  "feature": "<version>",
+  "timestamp": "<ISO-8601>",
+  "releaseType": "<hotfix|patch|minor|major>",
+  "verdict": "<GO|CONDITIONAL_GO|NO_GO>",
+  "gateScores": {"coverage": <n>, "bugStatus": <n>, "regression": <n>, "deployment": <n>},
+  "openBlockers": <n>,
+  "conditions": ["<condition1>"]
+}'
+
+node scripts/state-manager.js merge session findings '{
+  "area": "<version>",
+  "type": "release-readiness",
+  "summary": "Release assessment: <verdict>. Gates: Coverage <n>, Bugs <n>, Regression <n>, Deployment <n>. Blockers: <n>."
+}'
+```
 
 ## Output Structure
+
 ```
 ## Release Readiness Assessment
 Version: <version>
 Release Type: <type>
 Date: YYYY-MM-DD
-Assessor: QA Toolkit
 
 ### Overall Verdict: GO / NO-GO / CONDITIONAL GO
 Confidence: <percentage>
@@ -66,11 +145,22 @@ Confidence: <percentage>
 - <condition that triggers rollback>
 ```
 
-For release type templates, read `references/release-types.md`.
+## Output Section Order
+
+1. **Context preamble** — blockquote with project context and pre-filled signals note
+2. **Overall Verdict** — GO/CONDITIONAL GO/NO-GO with confidence score
+3. **Quality Gate Scores** — table with per-gate scores and status icons
+4. **Risk Matrix** — pre-populated from session state risks + new findings
+5. **Conditions** — if Conditional Go, actionable list
+6. **Post-Release Monitoring Plan**
+7. **Rollback Trigger Criteria**
+8. **Suggested Next Steps**
 
 ## Save
 Save to `qa-artifacts/release-assessments/release-YYYY-MM-DD-<version>.md`
 
 ## Suggested Next Steps
-After generating the assessment, suggest based on results:
-- If verdict is **NO-GO** or Regression Score is below 70: "Plan targeted regression testing with `/qa-toolkit:regression-planner`."
+After the assessment:
+- If verdict is **NO-GO** or Regression Score < 70: "Plan targeted regression testing with `/qa-toolkit:regression-planner`."
+- If Bug Status Score < 70: "File and triage remaining bugs with `/qa-toolkit:bug-report`."
+- If verdict is **GO** or **CONDITIONAL GO**: "Run a final risk prioritization with `/qa-toolkit:risk-prioritization` to confirm no missed areas."
diff --git a/skills/risk-prioritization/SKILL.md b/skills/risk-prioritization/SKILL.md
index d4db7a7..af53ab6 100644
--- a/skills/risk-prioritization/SKILL.md
+++ b/skills/risk-prioritization/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: risk-prioritization
-description: Prioritize test areas by risk using change frequency, business criticality, and defect density to allocate testing effort effectively.
+description: Rank test areas by composite risk score — change frequency, business criticality, and defect density — to allocate testing effort where it matters most. Use this skill when a user asks "where should we focus testing?", needs to prioritize a large test surface, or wants a data-driven testing strategy for a sprint or release cycle.
 ---
 
 # Risk-Based Test Prioritizer
diff --git a/skills/setup/SKILL.md b/skills/setup/SKILL.md
index 60698fa..a968b8d 100644
--- a/skills/setup/SKILL.md
+++ b/skills/setup/SKILL.md
@@ -1,35 +1,47 @@
 ---
 name: setup
-description: Read auto-detected project config and existing docs, confirm detection results with the user, and save corrections.
+description: Read auto-detected project config and existing docs, confirm detection results with the user, and save corrections and preferences. Use this skill when starting a new project with QA Toolkit, after installation, or when a user asks to check what was detected, configure QA settings, or update tracker or risk preferences.
 ---
 
 # QA Toolkit Setup
 
-Read `qa-artifacts/.qa-config.json` for auto-detected project context.
+Show the user what QA Toolkit detected about their project, read relevant existing docs for context, and save any corrections to project state.
 
 ## Workflow
 
-1. **Show the user what was detected** — languages, frameworks, test tools, CI/CD, package manager
-2. **Read existing docs** if found:
+1. **Read project state:**
+   ```bash
+   node scripts/state-manager.js read project detection
+   ```
+   Show the user what was detected: languages, frameworks, test tools, CI/CD, package manager, monorepo, test directories.
+
+2. **Read existing docs** if found (from detection):
    - `CLAUDE.md` — project conventions, existing guidelines
    - `README.md` — project overview
    - `TESTING.md` — testing standards (if exists)
    - `CONTRIBUTING.md` — contribution workflow (if exists)
+
 3. **Ask the user** to confirm or correct:
    - Is the detected tech stack correct?
    - What bug tracker does the team use? (Jira / GitHub Issues / Azure DevOps / Linear)
-   - Any high-risk areas of the app the team should focus on?
-4. **Update `qa-artifacts/.qa-config.json`** with corrections and preferences
+   - Any high-risk areas of the app the team should focus testing on?
+
+4. **Save corrections and preferences** using state-manager:
+   ```bash
+   node scripts/state-manager.js write project <key> <json-value>
+   ```
+   Save user-confirmed values (tracker preference, risk areas, corrections to detection).
+
 5. **Suggest adding output directory to `.gitignore`** if not already present:
-   - Check if `qa-artifacts/` (or the configured `outputDir`) is in the project's `.gitignore`
-   - If not, suggest adding it — QA artifacts contain machine-specific paths and session-specific output that shouldn't be committed
-   - Example line to add: `qa-artifacts/`
-6. **Summarize** what's configured and which skills are most relevant for this project
+   - `qa-artifacts/` contains machine-specific paths and session-specific output that shouldn't be committed
+   - If not already in `.gitignore`, suggest adding: `qa-artifacts/`
+
+6. **Summarize** what's configured and which skills are most relevant for this project's stack.
 
 ## Important
-- **Read only** — do not create or modify CLAUDE.md, README, or any existing project files
+- **Read only** — do not create or modify `CLAUDE.md`, `README.md`, or any existing project files
 - This is a configuration step, not a documentation generator
 
 ## Suggested Next Steps
-After setup is complete, suggest:
+After setup is complete, suggest based on the user's stated goals:
 - "Start with a PR review (`/qa-toolkit:pr-review`) or generate test cases from your requirements (`/qa-toolkit:test-cases`)."
diff --git a/skills/test-cases/SKILL.md b/skills/test-cases/SKILL.md
index fcff41b..5f9ba6c 100644
--- a/skills/test-cases/SKILL.md
+++ b/skills/test-cases/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: test-cases
-description: Generate state-aware test cases from requirements in table, Gherkin BDD, checklist, or JSON format with traceability IDs and inline risk annotations.
+description: Generate structured test cases from any requirements input — user stories, PRDs, acceptance criteria, Jira tickets, or vague feature descriptions — in table, Gherkin, checklist, or JSON format with traceability IDs. Use this skill whenever a user provides requirements and wants to know what to test, asks "write test cases for X", or shares a feature description, even informally.
 ---
 
 # Test Case Generator
diff --git a/skills/test-data/SKILL.md b/skills/test-data/SKILL.md
index 4702ba1..c2dd0d7 100644
--- a/skills/test-data/SKILL.md
+++ b/skills/test-data/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: test-data
-description: Generate realistic, privacy-safe synthetic test data with state-aware schema detection for JSON, CSV, or SQL output.
+description: Generate realistic, privacy-safe synthetic test data with schema-aware field generation for JSON, CSV, or SQL output. Use this skill whenever a user needs fake data for testing, asks for test fixtures, wants to seed a database, or needs data matching specific models — especially when they mention HIPAA/PII-safe requirements, specific record counts, or existing schema files.
 ---
 
 # Test Data Generator
diff --git a/skills/test-plan/SKILL.md b/skills/test-plan/SKILL.md
index 90fc30f..8bc1741 100644
--- a/skills/test-plan/SKILL.md
+++ b/skills/test-plan/SKILL.md
@@ -1,6 +1,6 @@
 ---
 name: test-plan
-description: Generate state-aware test plans from PRD or epic descriptions with scope, approach, environments, risks, and entry/exit criteria.
+description: Generate a comprehensive test plan from PRD, epic, or requirement input covering scope, approach, environments, risks, and entry/exit criteria. Use this skill when a user is starting a testing effort, planning for a feature or sprint, or asks "how should we test this?" or "we need a test plan for X" — before writing individual test cases.
 ---
 
 # Test Plan Generator

From 59dfdb9d25e9a88ee5d42cf115de5602346718b1 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Sat, 7 Mar 2026 05:21:42 +0530
Subject: [PATCH 13/15] docs: add workflow diagram, quick start guide, fix
 stale references
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Add Mermaid workflow diagram to README and GitHub Pages site
- Add Quick Start section with recommended first command and workflow
- Add agent selection guidance (when to use agents vs skills)
- Fix plugin.json description (10 → 15 skills)
- Fix CLAUDE.md skill count and stale save-artifact.js reference
- Fix CHANGELOG.md .qa-config.json → .qa-context.json references
- Fix artifact-organization.md save-artifact.js → session-hook.js

Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 .claude-plugin/plugin.json                    |  2 +-
 CHANGELOG.md                                  |  4 +-
 CLAUDE.md                                     |  4 +-
 README.md                                     | 37 ++++++++++++++++
 docs/index.html                               | 43 +++++++++++++++++++
 docs/style.css                                | 16 +++++++
 scripts/session-hook.js                       |  2 +-
 .../references/artifact-organization.md       |  2 +-
 8 files changed, 103 insertions(+), 7 deletions(-)

diff --git a/.claude-plugin/plugin.json b/.claude-plugin/plugin.json
index 180c8a0..78b6979 100644
--- a/.claude-plugin/plugin.json
+++ b/.claude-plugin/plugin.json
@@ -1,7 +1,7 @@
 {
   "name": "qa-toolkit",
   "version": "2.0.0",
-  "description": "Claude Code plugin for QA testers. Structured PR reviews, test case generation, bug reports, regression planning, and 6 more QA workflows. Auto-detects your tech stack. No code required.",
+  "description": "Claude Code plugin with 15 QA skills and 3 agent personas. PR reviews, test cases, bug reports, regression plans, coverage gaps, risk analysis, and more. Auto-detects your tech stack.",
   "author": {
     "name": "QA Skills"
   },
diff --git a/CHANGELOG.md b/CHANGELOG.md
index f3e5b89..739b62c 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,7 +5,7 @@
 QA Toolkit v2.0 is a ground-up rethink of how the plugin operates. The core idea: every skill now reads your project state before producing output, so artifacts build on each other rather than starting cold every time.
 
 ### Added
-- **State-aware skill system** — all skills read `.qa-config.json` on invocation, adapt output to detected stack, and write findings back so subsequent skills have context
+- **State-aware skill system** — all skills read `.qa-context.json` on invocation, adapt output to detected stack, and write findings back so subsequent skills have context
 - **Shared references** (`skills/shared/references/`) — `context-preamble.md`, `state-integration.md`, `output-formats.md`, `artifact-organization.md` loaded by all skills to enforce consistent behavior
 - **5 new skills**: `test-plan`, `exploratory-testing`, `coverage-gap`, `risk-prioritization`, `flaky-test-diagnosis`
 - **Multi-format output** — skills produce output in the format best suited to the artifact type (Markdown reports, Gherkin, cURL/Postman/Playwright, JSON/CSV/SQL, etc.)
@@ -37,7 +37,7 @@ QA Toolkit v2.0 is a ground-up rethink of how the plugin operates. The core idea
 - `hooksEnabled` toggle to disable hooks without removing the plugin
 - 27 automated tests using Node.js built-in test runner (zero dependencies)
 - GitHub Actions CI workflow running tests on Node 18/20/22
-- `.qa-config.json` schema documentation in README
+- `.qa-context.json` schema documentation in README
 - `.gitignore` suggestion step in setup skill workflow
 - "Permissions & Side Effects" section in README with `Bash(git:*)` mechanism explained
 - Contextual "Suggested Next Steps" cross-references in all skills
diff --git a/CLAUDE.md b/CLAUDE.md
index 4ed2970..f3c7516 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -2,7 +2,7 @@
 
 ## What This Is
 
-A Claude Code plugin providing 10 QA skills (slash commands) and 3 agent personas. Auto-detects project tech stack and produces structured QA artifacts.
+A Claude Code plugin providing 15 QA skills (slash commands) and 3 agent personas. Auto-detects project tech stack and produces structured QA artifacts.
 
 ## Project Structure
 
@@ -11,7 +11,7 @@ A Claude Code plugin providing 10 QA skills (slash commands) and 3 agent persona
 settings.json               — Default agent and plugin settings
 hooks/hooks.json            — Event hooks (SessionStart, Stop)
 scripts/detect-project.js   — Auto-detection script (runs on SessionStart)
-scripts/save-artifact.js    — Activity logger (runs on Stop)
+scripts/session-hook.js     — State promotion, archival, activity logger (runs on Stop)
 agents/<name>.md            — Agent persona definitions (conversational, multi-turn)
 skills/<name>/SKILL.md      — Skill definitions (one-shot structured output)
 skills/<name>/references/   — Domain-specific knowledge files
diff --git a/README.md b/README.md
index 21d896d..bbf9882 100644
--- a/README.md
+++ b/README.md
@@ -32,6 +32,41 @@ So I built QA Toolkit. It auto-detects your stack, remembers your project contex
 
 ---
 
+## Quick Start
+
+```bash
+# 1. Install
+claude plugin add github:cyberwalk3r/qa-toolkit
+
+# 2. Open any project — detection runs automatically
+
+# 3. Confirm your setup (recommended first command)
+/qa-toolkit:setup
+```
+
+After setup, try `/qa-toolkit:pr-review` on a real PR or `/qa-toolkit:test-cases` with a feature description. Each skill output includes **Suggested Next Steps** pointing you to related commands.
+
+All output lands in `qa-artifacts/` — add it to `.gitignore` if you don't want to commit QA artifacts.
+
+### Recommended Workflow
+
+Skills are most powerful when chained. Each one feeds context into the next:
+
+```mermaid
+graph LR
+    setup --> pr-review
+    pr-review --> regression-planner
+    regression-planner --> release-readiness
+    pr-review --> bug-report
+    regression-planner --> test-cases
+    test-cases --> e2e-test
+    test-cases --> test-data
+```
+
+Start anywhere — every skill works standalone — but chaining builds richer context.
+
+---
+
 ## How It Works
 
 **1. Install the plugin**
@@ -105,6 +140,8 @@ Agents remember context across the conversation. The more you share, the sharper
 
 **Agents** are **conversational** — use them for ongoing discussion, decision-making, and exploratory work.
 
+> **When to pick an agent:** Use `qa-reviewer` when walking through a PR interactively. Use `qa-explorer` when brainstorming edge cases for a feature. Use `qa-lead` when deciding whether a release is ready to ship. If you want a saved artifact instead of a conversation, use the corresponding skill.
+
 ---
 
 ## What's New in v2.0
diff --git a/docs/index.html b/docs/index.html
index 2687cf2..eb4ee4b 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -60,6 +60,23 @@
   <link href="https://fonts.googleapis.com/css2?family=Syne:wght@600;700;800&family=Inter:wght@400;500;600&family=JetBrains+Mono:wght@400;700&display=swap" rel="stylesheet">
 
   <link rel="stylesheet" href="style.css">
+  <script type="module">
+    import mermaid from 'https://cdn.jsdelivr.net/npm/mermaid@11/dist/mermaid.esm.min.mjs';
+    mermaid.initialize({
+      startOnLoad: true,
+      theme: 'dark',
+      themeVariables: {
+        primaryColor: '#1e1e1e',
+        primaryTextColor: '#e5e7eb',
+        primaryBorderColor: '#f59e0b',
+        lineColor: '#f59e0b',
+        secondaryColor: '#171717',
+        tertiaryColor: '#111111',
+        fontFamily: 'Inter, sans-serif',
+        fontSize: '14px'
+      }
+    });
+  </script>
 </head>
 <body>
 
@@ -73,6 +90,7 @@
   <span class="logo">/qa-toolkit</span>
   <div class="nav-links">
     <a href="#how-it-works">How It Works</a>
+    <a href="#workflow">Workflow</a>
     <a href="#commands">Commands</a>
     <a href="#agents">Agents</a>
     <a href="#install">Install</a>
@@ -191,6 +209,31 @@ <h3>Describe what you need</h3>
   </div>
 </section>
 
+<!-- Workflow -->
+<section id="workflow" class="workflow-section">
+  <div class="shimmer-line" style="top:0;" aria-hidden="true"></div>
+  <div class="container">
+    <div data-animate-slide>
+      <div class="section-label">Workflow</div>
+      <h2>Skills chain together. Each one feeds the next.</h2>
+      <p class="section-sub">Start anywhere — every command works standalone. But chaining builds richer, state-aware context.</p>
+    </div>
+    <div class="mermaid-wrapper" data-animate-scale>
+      <pre class="mermaid">
+graph LR
+    setup --> pr-review
+    pr-review --> regression-planner
+    regression-planner --> release-readiness
+    pr-review --> bug-report
+    regression-planner --> test-cases
+    test-cases --> e2e-test
+    test-cases --> test-data
+      </pre>
+    </div>
+    <p class="install-note" data-animate>Run <code class="inline-code">/qa-toolkit:setup</code> first to confirm your detected stack, then follow the <strong>Suggested Next Steps</strong> at the bottom of each artifact.</p>
+  </div>
+</section>
+
 <!-- Commands -->
 <section id="commands" class="commands-section">
   <!-- Ambient orbs -->
diff --git a/docs/style.css b/docs/style.css
index c30abf5..63acb50 100644
--- a/docs/style.css
+++ b/docs/style.css
@@ -756,6 +756,22 @@ h2 {
 }
 
 /* ===================== COMMANDS ===================== */
+.workflow-section {
+  border-top: 1px solid var(--border);
+}
+.mermaid-wrapper {
+  background: var(--code-bg);
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  padding: 2rem 1.5rem;
+  margin-bottom: 1.25rem;
+  overflow-x: auto;
+  transition: border-color 0.3s ease, box-shadow 0.3s ease;
+}
+.mermaid-wrapper:hover {
+  border-color: rgba(245,158,11,0.2);
+}
+
 .commands-section {
   background: var(--surface);
   border-top: 1px solid var(--border);
diff --git a/scripts/session-hook.js b/scripts/session-hook.js
index 00468a4..a08e817 100755
--- a/scripts/session-hook.js
+++ b/scripts/session-hook.js
@@ -5,7 +5,7 @@
  *
  * Runs on Stop to promote session findings to project state,
  * archive the completed session, and log activity. Replaces
- * save-artifact.js with expanded state-aware functionality.
+ * the original save-artifact.js with expanded state-aware functionality.
  *
  * Each section (promote, archive, log) is independent -- a failure
  * in one does not prevent the others from running.
diff --git a/skills/shared/references/artifact-organization.md b/skills/shared/references/artifact-organization.md
index e447faa..35ac9b8 100644
--- a/skills/shared/references/artifact-organization.md
+++ b/skills/shared/references/artifact-organization.md
@@ -91,7 +91,7 @@ Existing flat artifact directories (`qa-artifacts/test-cases/`, `qa-artifacts/bu
 
 - Old artifacts stay where they are
 - New artifacts use context-based paths when context is available
-- The `save-artifact.js` hook scans recursively, detecting artifacts in both flat and context-prefixed structures
+- The `session-hook.js` Stop hook scans recursively, detecting artifacts in both flat and context-prefixed structures
 - No cleanup or migration scripts are needed
 
 ## Integration with State Manager

From 8e8a7aa9914b34db63449db05338c78a013cd157 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Sat, 7 Mar 2026 05:50:15 +0530
Subject: [PATCH 14/15] docs: fix permissions docs, remove shipped .claude/,
 add dry-run and budget docs
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Remove .claude/settings.local.json — plugins should not ship user-local
  settings; permissions are approved interactively on first use
- Rewrite README Permissions section to describe interactive approval
- Document --dry-run flag for detect-project.js in README and CONTRIBUTING
- Document skill budget validator in CONTRIBUTING and add to CI
- Add session end summary output to session-hook.js
- Move test file from scripts/ to tests/
- Add CHANGELOG entry for v2.0.1

Signed-off-by: nerd <nerd@localhost>
Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 .claude/settings.local.json              | 17 --------
 .github/workflows/test.yml               |  1 +
 .gitignore                               |  3 ++
 CHANGELOG.md                             | 15 +++++++
 CONTRIBUTING.md                          | 18 +++++++-
 README.md                                | 48 ++++++++++++++++++---
 scripts/detect-project.js                | 54 +++++++++++++-----------
 scripts/session-hook.js                  |  7 +++
 {scripts => tests}/state-manager.test.js |  2 +-
 9 files changed, 116 insertions(+), 49 deletions(-)
 delete mode 100644 .claude/settings.local.json
 rename {scripts => tests}/state-manager.test.js (99%)

diff --git a/.claude/settings.local.json b/.claude/settings.local.json
deleted file mode 100644
index 02f631d..0000000
--- a/.claude/settings.local.json
+++ /dev/null
@@ -1,17 +0,0 @@
-{
-  "permissions": {
-    "allow": [
-      "Bash(git diff:*)",
-      "Bash(git log:*)",
-      "Bash(git show:*)",
-      "Bash(git branch:*)",
-      "Bash(git status:*)",
-      "Bash(git fetch:*)",
-      "Bash(node scripts/detect-project.js:*)",
-      "Bash(node scripts/session-hook.js:*)",
-      "Bash(node scripts/state-manager.js:*)",
-      "Bash(npm install:*)",
-      "Bash(node:*)"
-    ]
-  }
-}
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
index 51b220a..2e844a6 100644
--- a/.github/workflows/test.yml
+++ b/.github/workflows/test.yml
@@ -18,3 +18,4 @@ jobs:
         with:
           node-version: ${{ matrix.node-version }}
       - run: node --test tests/*.test.js
+      - run: node scripts/validate-skill-budget.js
diff --git a/.gitignore b/.gitignore
index 3e57f06..e4ea5d2 100644
--- a/.gitignore
+++ b/.gitignore
@@ -15,6 +15,9 @@ Thumbs.db
 desktop.ini
 ehthumbs.db
 
+# Claude Code user settings (not part of the plugin)
+.claude/
+
 # Generated at runtime in target projects
 qa-artifacts/
 
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 739b62c..70af2d0 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,20 @@
 # Changelog
 
+## [2.0.1] — 2026-03-07
+
+### Added
+- `--dry-run` flag for `detect-project.js` — preview detection without writing files
+- Session end summary output showing promoted findings and modified artifact count
+- Skill budget validation (`validate-skill-budget.js`) added to CI pipeline
+- GitHub Pages documentation site (`docs/`)
+
+### Changed
+- Removed `.claude/settings.local.json` — plugins should not ship user-local settings; permissions are approved interactively on first use
+- Moved test files from `scripts/` to `tests/` directory
+
+### Fixed
+- README "Permissions & Side Effects" section now accurately describes interactive approval (not pre-approved settings)
+
 ## [2.0.0] — 2026-03-06
 
 QA Toolkit v2.0 is a ground-up rethink of how the plugin operates. The core idea: every skill now reads your project state before producing output, so artifacts build on each other rather than starting cold every time.
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index bdbb02a..16feeec 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -77,10 +77,18 @@ agents/<agent-name>.md        # Required: agent definition with frontmatter
 ## Testing
 
 ```bash
+# Run tests
 node --test tests/*.test.js
+
+# Validate skill registration budget (must stay under 16K chars)
+node scripts/validate-skill-budget.js
 ```
 
-Tests use Node.js built-in test runner (zero dependencies). When adding a skill or modifying a script, add or update the corresponding test in `tests/`.
+Both run in CI on Node 18/20/22. Tests use Node.js built-in test runner (zero dependencies). When adding a skill or modifying a script, add or update the corresponding test in `tests/`.
+
+### Skill budget
+
+The combined character count of all skill `name` + `description` frontmatter fields must stay under 16,000 characters (Claude Code registration limit). Run `node scripts/validate-skill-budget.js` to check. The `--json` flag produces machine-readable output for CI.
 
 ## Local Development
 
@@ -90,6 +98,14 @@ claude --debug --plugin-dir ./qa-toolkit
 
 The `--debug` flag shows hook execution and skill loading in the Claude Code output.
 
+### Preview detection
+
+To see what the SessionStart hook would detect without writing any files:
+
+```bash
+node scripts/detect-project.js --dry-run
+```
+
 ---
 
 Something missing or wrong? Open an issue — this doc is meant to stay short, not stay incomplete.
diff --git a/README.md b/README.md
index bbf9882..94b3146 100644
--- a/README.md
+++ b/README.md
@@ -198,9 +198,40 @@ qa-artifacts/
 
 ---
 
+## Advanced: State Management
+
+Skills and hooks use `scripts/state-manager.js` to read and write QA state. You can also use it directly to inspect or modify state.
+
+```bash
+# Read full project or session state
+node scripts/state-manager.js read project
+node scripts/state-manager.js read session
+
+# Read a specific field
+node scripts/state-manager.js read project risks
+node scripts/state-manager.js read session featureUnderTest
+
+# Write a field (replaces value; appends for session findings/skillHistory)
+node scripts/state-manager.js write session featureUnderTest '"login flow"'
+node scripts/state-manager.js write session findings '{"type":"bug","area":"auth"}'
+
+# Merge into an array field (deduplicates by key)
+node scripts/state-manager.js merge project risks '{"area":"auth","severity":"high"}'
+
+# Initialize a fresh session (archives existing if it has content)
+node scripts/state-manager.js init session
+
+# Archive current session to qa-artifacts/sessions/
+node scripts/state-manager.js archive session
+```
+
+State files: `qa-artifacts/.qa-context.json` (project), `qa-artifacts/.qa-session.json` (session). Both use atomic writes and schema versioning.
+
+---
+
 ## Permissions & Side Effects
 
-The plugin ships a `.claude/settings.local.json` that pre-approves the following permissions:
+On first use, Claude Code will prompt you to approve each permission the plugin needs. You can approve, deny, or revoke permissions at any time. The plugin requires:
 
 | Permission | Used by |
 |------------|---------|
@@ -214,15 +245,20 @@ The plugin ships a `.claude/settings.local.json` that pre-approves the following
 | `Bash(node scripts/session-hook.js:*)` | Stop hook — session archiving and activity log |
 | `Bash(node scripts/state-manager.js:*)` | All skills — reading and writing QA state |
 
-You'll see these listed when you install the plugin and can revoke them at any time.
+**No network access.** All scripts use Node.js stdlib only — zero external dependencies, zero network calls.
 
 **Automatic behavior:**
-- **Session start:** Scans project marker files (languages, frameworks, test tools, CI/CD), reads the first 50 lines of `CLAUDE.md` if present (to detect project conventions), writes `qa-artifacts/.qa-context.json`. No network calls.
-- **Session end:** Promotes session findings to project state, archives the session, logs which artifacts were created/modified to `qa-artifacts/.qa-activity.log`. No network calls.
+- **Session start:** Scans project marker files (languages, frameworks, test tools, CI/CD), reads the first 50 lines of `CLAUDE.md` if present (to detect project conventions), writes `qa-artifacts/.qa-context.json`.
+- **Session end:** Promotes session findings to project state, archives the session, logs which artifacts were created/modified to `qa-artifacts/.qa-activity.log`.
 
 **Output directory:** All artifacts are written to `qa-artifacts/` in your project root. This directory is runtime-generated and should be gitignored — add `qa-artifacts/` to your project's `.gitignore` if you don't want to commit QA artifacts.
 
-**To disable hooks** without removing the plugin:
+**Preview detection without writing files:**
+```bash
+node scripts/detect-project.js --dry-run
+```
+
+**To disable hooks** without removing the plugin, set `hooksEnabled` to `false` in `settings.json`:
 ```json
 {
     "agent": "qa-reviewer",
@@ -243,7 +279,7 @@ claude plugin add github:cyberwalk3r/qa-toolkit
 claude plugin add ./qa-toolkit
 ```
 
-The plugin ships with `.claude/settings.local.json` that pre-approves its own hook scripts and read-only git commands so everything works on first use without confirmation prompts. See the [Permissions & Side Effects](#permissions--side-effects) section for the full list.
+On first use, Claude Code will prompt you to approve the plugin's hook scripts and read-only git commands. See the [Permissions & Side Effects](#permissions--side-effects) section for the full list.
 
 ---
 
diff --git a/scripts/detect-project.js b/scripts/detect-project.js
index 3f14844..b754109 100644
--- a/scripts/detect-project.js
+++ b/scripts/detect-project.js
@@ -753,35 +753,41 @@ projectState.detection = detection;
 projectState.conventions = conventions;
 projectState.detectionTiming = detectionTiming;
 
-// Save to .qa-context.json (atomic write)
-if (!fs.existsSync(configDir)) {
-    fs.mkdirSync(configDir, { recursive: true });
+const dryRun = process.argv.includes('--dry-run');
+
+if (!dryRun) {
+    // Save to .qa-context.json (atomic write)
+    if (!fs.existsSync(configDir)) {
+        fs.mkdirSync(configDir, { recursive: true });
+    }
+    const tmpContextPath = contextPath + '.tmp';
+    fs.writeFileSync(tmpContextPath, JSON.stringify(projectState, null, 2));
+    fs.renameSync(tmpContextPath, contextPath);
 }
-const tmpContextPath = contextPath + '.tmp';
-fs.writeFileSync(tmpContextPath, JSON.stringify(projectState, null, 2));
-fs.renameSync(tmpContextPath, contextPath);
 
 // ---- Initialize session state ----
 
-let sessionStatus = 'new (first session)';
-try {
-    execFileSync('node', [stateManagerPath, 'init', 'session'], {
-        cwd: projectRoot,
-        env: { ...process.env, CLAUDE_PLUGIN_ROOT: pluginRoot },
-        encoding: 'utf8',
-        timeout: 5000,
-    });
-    const sessionsDir = path.join(configDir, 'sessions');
-    if (fs.existsSync(sessionsDir)) {
-        try {
-            const archived = fs.readdirSync(sessionsDir).filter(f => f.endsWith('.json'));
-            if (archived.length > 0) {
-                sessionStatus = 'new (previous archived)';
-            }
-        } catch { /* ignore */ }
+let sessionStatus = dryRun ? 'dry-run (no writes)' : 'new (first session)';
+if (!dryRun) {
+    try {
+        execFileSync('node', [stateManagerPath, 'init', 'session'], {
+            cwd: projectRoot,
+            env: { ...process.env, CLAUDE_PLUGIN_ROOT: pluginRoot },
+            encoding: 'utf8',
+            timeout: 5000,
+        });
+        const sessionsDir = path.join(configDir, 'sessions');
+        if (fs.existsSync(sessionsDir)) {
+            try {
+                const archived = fs.readdirSync(sessionsDir).filter(f => f.endsWith('.json'));
+                if (archived.length > 0) {
+                    sessionStatus = 'new (previous archived)';
+                }
+            } catch { /* ignore */ }
+        }
+    } catch (err) {
+        sessionStatus = 'init failed (non-blocking)';
     }
-} catch (err) {
-    sessionStatus = 'init failed (non-blocking)';
 }
 
 // ---- Output context ----
diff --git a/scripts/session-hook.js b/scripts/session-hook.js
index a08e817..711ad47 100755
--- a/scripts/session-hook.js
+++ b/scripts/session-hook.js
@@ -170,3 +170,10 @@ try {
 } catch (err) {
     console.error(`Failed to log activity: ${err.message}`);
 }
+
+// ---- Summary output ----
+
+const summaryParts = ['[QA Toolkit] Session ended.'];
+if (findingsCount > 0) summaryParts.push(`Findings promoted: ${findingsCount}`);
+if (typeof recentFiles !== 'undefined' && recentFiles.length > 0) summaryParts.push(`Artifacts modified: ${recentFiles.length}`);
+console.log(summaryParts.join(' '));
diff --git a/scripts/state-manager.test.js b/tests/state-manager.test.js
similarity index 99%
rename from scripts/state-manager.test.js
rename to tests/state-manager.test.js
index 64785c0..a4a1b94 100644
--- a/scripts/state-manager.test.js
+++ b/tests/state-manager.test.js
@@ -12,7 +12,7 @@ const fs = require('fs');
 const path = require('path');
 const assert = require('assert');
 
-const SCRIPT = path.join(__dirname, 'state-manager.js');
+const SCRIPT = path.join(__dirname, '..', 'scripts', 'state-manager.js');
 const TEST_DIR = path.join(__dirname, '..', '__test_tmp__');
 const QA_DIR = path.join(TEST_DIR, 'qa-artifacts');
 const PROJECT_STATE = path.join(QA_DIR, '.qa-context.json');

From 5d848be795de9fa3d084fbc651f1e1259face247 Mon Sep 17 00:00:00 2001
From: Himanshu Vimal <himanshu.vimal@gmail.com>
Date: Sat, 7 Mar 2026 06:08:30 +0530
Subject: [PATCH 15/15] fix: stale docs, log rotation, author metadata
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Fix artifact directory names in README (coverage-gaps → coverage-analysis,
  risk-analyses → risk-analysis, flaky-diagnoses → flaky-diagnosis, add
  missing exploratory/)
- Fix .qa-context.json schema table to show actual nested structure
  (detection.* sub-fields, not flat top-level)
- Add settings.json schema table and detection limitations section
- Add activity log rotation at 512 KB in session-hook.js
- Update CLAUDE.md skill description convention to match reality (1-3
  sentences, not one)
- Fix author metadata to cyberwalk3r in plugin.json and LICENSE

Signed-off-by: nerd <nerd@localhost>
Signed-off-by: Himanshu Vimal <himanshu.vimal@gmail.com>
---
 .claude-plugin/plugin.json |  2 +-
 CLAUDE.md                  |  2 +-
 LICENSE                    |  2 +-
 README.md                  | 95 ++++++++++++++++++++++++--------------
 scripts/session-hook.js    | 17 +++++++
 5 files changed, 81 insertions(+), 37 deletions(-)

diff --git a/.claude-plugin/plugin.json b/.claude-plugin/plugin.json
index 78b6979..a1005a9 100644
--- a/.claude-plugin/plugin.json
+++ b/.claude-plugin/plugin.json
@@ -3,7 +3,7 @@
   "version": "2.0.0",
   "description": "Claude Code plugin with 15 QA skills and 3 agent personas. PR reviews, test cases, bug reports, regression plans, coverage gaps, risk analysis, and more. Auto-detects your tech stack.",
   "author": {
-    "name": "QA Skills"
+    "name": "cyberwalk3r"
   },
   "repository": {
     "type": "git",
diff --git a/CLAUDE.md b/CLAUDE.md
index f3c7516..7f6b1a3 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -21,7 +21,7 @@ skills/<name>/scripts/      — Helper scripts (e.g., Playwright runner)
 ## Conventions
 
 ### Skills
-- `SKILL.md` must have YAML frontmatter with `name` and `description` (one sentence)
+- `SKILL.md` must have YAML frontmatter with `name` and `description` (1-3 sentences covering what it does and when to trigger it)
 - Skills produce one-shot structured output and save to the configured output directory
 - Each skill reads `<outputDir>/.qa-context.json` for project context
 - Reference files contain non-obvious, specific guidance — not generic info
diff --git a/LICENSE b/LICENSE
index fd983f9..74d01df 100644
--- a/LICENSE
+++ b/LICENSE
@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2026 QA Skills
+Copyright (c) 2026 cyberwalk3r
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal
diff --git a/README.md b/README.md
index 94b3146..387c60d 100644
--- a/README.md
+++ b/README.md
@@ -22,13 +22,11 @@ claude plugin add github:cyberwalk3r/qa-toolkit
 
 ---
 
-## Why I Built This
+## Motivation
 
 QA work is repetitive in the worst way. Every bug report needs the same structure. Every PR review needs the same checks. Every regression plan needs the same risk analysis. And every team member does it differently.
 
-I wanted a tool that handles the formatting so you can focus on the judgment. Describe what you're testing in plain English — get back something you can actually use.
-
-So I built QA Toolkit. It auto-detects your stack, remembers your project context, and produces consistent QA artifacts every time. Whether you're a dedicated QA lead or a developer wearing the QA hat, it adapts to your stack and your tracker without asking you to configure anything.
+QA Toolkit handles the formatting so you can focus on the judgment. Describe what you're testing in plain English — get back something you can actually use. It auto-detects your stack, remembers your project context, and produces consistent QA artifacts every time. Whether you're a dedicated QA lead or a developer wearing the QA hat, it adapts to your stack and your tracker without asking you to configure anything.
 
 ---
 
@@ -144,55 +142,73 @@ Agents remember context across the conversation. The more you share, the sharper
 
 ---
 
-## What's New in v2.0
+## Output
 
-**State-aware skills.** Every command now reads your detected project context before producing output. It knows you're using Jest, not Pytest. It knows your CI is GitHub Actions. It knows what artifacts you've already generated this session.
+Everything saves to `qa-artifacts/` in your project root (configurable via `settings.json`).
 
-**15 skills** (up from 10). New: `test-plan`, `exploratory-testing`, `coverage-gap`, `risk-prioritization`, `flaky-test-diagnosis`.
+### settings.json
 
-**Redesigned agents.** `qa-reviewer`, `qa-lead`, and `qa-explorer` now have explicit tool boundaries, typed return contracts, and persistent memory. Their behavior is predictable and composable.
+| Setting | Type | Default | Description |
+|---------|------|---------|-------------|
+| `agent` | string | `"qa-reviewer"` | Default agent persona for interactive sessions |
+| `outputDir` | string | `"qa-artifacts"` | Output directory (relative path, no `..`) |
+| `hooksEnabled` | boolean | `true` | Enable/disable SessionStart and Stop hooks |
+| `detectionBudgetMs` | number | `500` | Time budget (ms) for project auto-detection phases |
 
----
-
-## Output
-
-Everything saves to `qa-artifacts/` in your project root (configurable via `settings.json`):
+### Artifact structure
 
 ```
 qa-artifacts/
 ├── .qa-context.json         # Auto-detected project config and session state
 ├── .qa-activity.log         # Session activity log
-├── pr-reviews/
-├── bug-reports/
-├── test-cases/
-├── test-plans/
+├── a11y-audits/
 ├── api-tests/
+├── bug-reports/
+├── coverage-analysis/
+├── e2e-tests/
+├── exploratory/
+├── flaky-diagnosis/
+├── pr-reviews/
 ├── regression-plans/
-├── test-data/
-├── a11y-audits/
 ├── release-assessments/
-├── e2e-tests/
-├── coverage-gaps/
-├── risk-analyses/
-└── flaky-diagnoses/
+├── risk-analysis/
+├── test-cases/
+├── test-data/
+└── test-plans/
 ```
 
 ### .qa-context.json Schema
 
+Top-level fields:
+
 | Field | Type | Description |
 |-------|------|-------------|
-| `detectedAt` | string | ISO 8601 timestamp |
+| `schemaVersion` | number | Schema version (currently `1`) |
+| `detectedAt` | string | ISO 8601 timestamp of last detection |
 | `projectRoot` | string | Absolute path to project directory |
 | `outputDir` | string | Configured output directory name |
-| `languages` | string[] | Detected languages |
-| `frameworks` | string[] | Detected frameworks |
-| `testFrameworks` | string[] | Detected test tools |
-| `cicd` | string[] | Detected CI/CD systems |
-| `packageManager` | string\|null | Detected package manager |
-| `hasClaudeMd` | boolean | Whether CLAUDE.md exists |
-| `hasReadme` | boolean | Whether README.md exists |
-| `existingDocs` | object[] | Detected documentation files |
-| `existingTestDirs` | string[] | Detected test directories |
+| `detection` | object | Auto-detected project info (see below) |
+| `conventions` | object | Formatting, linting, TypeScript, test patterns |
+| `detectionTiming` | object | Phase timing and budget data |
+| `risks` | object[] | Accumulated risk findings across sessions |
+| `coverageGaps` | object[] | Accumulated coverage gap findings |
+| `testingConventions` | object[] | Project testing conventions |
+| `historicalFindings` | object | Session counts, bugs reported, areas reviewed |
+
+`detection` sub-fields:
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `detection.languages` | string[] | Detected languages |
+| `detection.frameworks` | string[] | Detected frameworks |
+| `detection.testFrameworks` | string[] | Detected test tools |
+| `detection.cicd` | string[] | Detected CI/CD systems |
+| `detection.packageManager` | string\|null | Detected package manager |
+| `detection.hasClaudeMd` | boolean | Whether CLAUDE.md exists |
+| `detection.hasReadme` | boolean | Whether README.md exists |
+| `detection.existingDocs` | object[] | Detected documentation files |
+| `detection.existingTestDirs` | string[] | Detected test directories |
+| `detection.monorepo` | object\|null | Workspace tool, config file, packages |
 
 **Note:** `projectRoot` is machine-specific — don't commit `.qa-context.json` to version control. Delete it to force re-detection.
 
@@ -244,11 +260,12 @@ On first use, Claude Code will prompt you to approve each permission the plugin
 | `Bash(node scripts/detect-project.js:*)` | SessionStart hook — project detection |
 | `Bash(node scripts/session-hook.js:*)` | Stop hook — session archiving and activity log |
 | `Bash(node scripts/state-manager.js:*)` | All skills — reading and writing QA state |
+| `Bash(npx playwright test:*)` | e2e-test skill — runs generated Playwright tests (user-initiated only) |
 
 **No network access.** All scripts use Node.js stdlib only — zero external dependencies, zero network calls.
 
 **Automatic behavior:**
-- **Session start:** Scans project marker files (languages, frameworks, test tools, CI/CD), reads the first 50 lines of `CLAUDE.md` if present (to detect project conventions), writes `qa-artifacts/.qa-context.json`.
+- **Session start:** Scans project marker files (languages, frameworks, test tools, CI/CD), writes `qa-artifacts/.qa-context.json`. If `CLAUDE.md` exists, reads the first 50 lines to detect project conventions (stored in `.qa-context.json` as `detection.existingQaConfig.claudeMdSummary` — no content is sent externally).
 - **Session end:** Promotes session findings to project state, archives the session, logs which artifacts were created/modified to `qa-artifacts/.qa-activity.log`.
 
 **Output directory:** All artifacts are written to `qa-artifacts/` in your project root. This directory is runtime-generated and should be gitignored — add `qa-artifacts/` to your project's `.gitignore` if you don't want to commit QA artifacts.
@@ -283,6 +300,16 @@ On first use, Claude Code will prompt you to approve the plugin's hook scripts a
 
 ---
 
+## Detection Limitations
+
+Auto-detection uses lightweight marker-file scanning with a configurable time budget (default 500ms). Phases that exceed the budget are skipped gracefully. Known limitations:
+
+- **Workspace config parsing** (`pnpm-workspace.yaml`, `Cargo.toml` workspaces, `go.work`, `pyproject.toml`) uses simplified regex-based parsers. Unusual formatting (inline comments, multi-line strings) may not be recognized. If detection misses your monorepo layout, run `/qa-toolkit:setup` to correct it manually.
+- **Glob-based language detection** (e.g., `*.csproj`) scans up to 3 directory levels deep. Deeply nested projects may not be detected.
+- **Activity log rotation** triggers at 512 KB, keeping the most recent half of entries. Long-running projects won't accumulate unbounded log files.
+
+---
+
 ## Works With Your Existing Setup
 
 - **Respects existing CLAUDE.md** — reads it for context, never overwrites
diff --git a/scripts/session-hook.js b/scripts/session-hook.js
index 711ad47..06027f3 100755
--- a/scripts/session-hook.js
+++ b/scripts/session-hook.js
@@ -121,11 +121,28 @@ try {
 
 // ---- 3. Log activity (absorbs save-artifact.js functionality) ----
 
+const MAX_LOG_BYTES = 512 * 1024; // 512 KB
+
 try {
     if (!fs.existsSync(configDir)) {
         fs.mkdirSync(configDir, { recursive: true });
     }
 
+    // Rotate log if it exceeds size cap
+    try {
+        if (fs.existsSync(logPath)) {
+            const stat = fs.statSync(logPath);
+            if (stat.size > MAX_LOG_BYTES) {
+                const content = fs.readFileSync(logPath, 'utf8');
+                const lines = content.split('\n');
+                // Keep the most recent half of lines
+                const keepFrom = Math.floor(lines.length / 2);
+                const rotated = '--- log rotated ---\n' + lines.slice(keepFrom).join('\n');
+                fs.writeFileSync(logPath, rotated);
+            }
+        }
+    } catch { /* rotation failure is non-blocking */ }
+
     // Scan for recently modified artifacts (last 5 minutes)
     const recentFiles = [];
     const cutoff = Date.now() - 5 * 60 * 1000;