P0–P2 polish: badges, architecture diagram, output mock-up, RU mirror, CI

[CreatmanCEO]

Summary

Single PR addressing the README/structure gap identified in deep comparison with the Hydrowatch standard. The actual workflow logic was already strong (clear problem statement, raw-MCP-vs-commands table, 41 K-message Telegram production test). The README was understating it: no architecture diagram, no output preview, no CI, no Limitations, no cross-link to the three sister repos.

What changed

P0 — credibility & accuracy

• Hero badges — License, Stars, Validate CI, Built on notebooklm-mcp-cli, Claude Code Opus 4.7, MCP, platforms
• Hero blurb leads with concrete production proof (7 commands · 41 K-message Telegram forum tested · 30+ frameworks)
• Flagship value-prop quote elevated to a callout under the hero
• Topics + description — applied via gh api after merge

P1 — structural strengthening

• docs/architecture.svg — pipeline diagram. One slash command drives a deterministic three-phase recipe (Collect → Analyse → Artifacts) over notebooklm-mcp-cli MCP tools.
• docs/output-mockup.svg — visual mock-up of /research output: structured findings, patterns, contradictions with inline citations.
• "What /research returns" section anchored by the mock-up.
• "Measured impact" table — 5 production scenarios with before/after numbers (research pipeline, 41 K-message Telegram, YouTube 429 workaround, 30+ frameworks, daily auth check).
• "When to use which research command" decision helper.
• "Limitations" section — cookie expiry rhythm, 500 K word cap, /edit-source workaround caveat, forum heuristic, opaque NotebookLM rate limits, Claude Code only, Windows-only notifications, upstream MCP dependency.
• "Related" cross-links to all three sister repos: claude-code-antiregression-setup, ai-context-hierarchy, claude-statusline.
• Project structure tree now matches the actual filesystem.
• CLAUDE.md for this repo — Level 1 file documenting architecture, CRITICAL RULES, commands, patterns. Pairs with the ai-context-hierarchy pattern.
• CHANGELOG.md (Keep a Changelog) starting at 0.1.0 → 0.2.0.
• CONTRIBUTING.md with priority list (Linux/macOS native notifications, expanded URL dictionary, new slash commands, locale translations, chunker improvements for WhatsApp/Discord/Slack).
• .github/workflows/validate.yml — bash -n, ShellCheck error severity, Python compile, command frontmatter check, SVG well-formed XML, every docs/* asset referenced from README exists, internal Markdown links resolve.

P2 — content

• README.ru.md fully mirrored to the new EN structure with the same badges, diagrams, Limitations, Measured impact, Related, decision helper.
• Author signature expanded with Habr / dev.to profile links.

What is NOT in this PR

• Habr / dev.to companion article. No NotebookLM-specific article exists yet; documented in CHANGELOG.md notes as the next traffic-driver.
• Issue / PR to jacob-bd/notebooklm-mcp-cli proposing reciprocal link. Tracked separately.
• awesome-claude-code submission — that channel currently blocked for non-collaborators across the maintainer's repo.

Test plan

• bash -n clean on scripts/nlm-auth-check.sh
• python -m py_compile clean on scripts/telegram-chunker.py
• Both new SVGs are well-formed XML
• Every commands/*.md has a # /command heading on first line
• All docs/* assets referenced from README and README.ru.md exist
• All internal Markdown links resolve
• LICENSE and CHANGELOG.md present
• ShellCheck (CI runs this — not installed locally)

🤖 Generated with Claude Code