Skip to content

Latest commit

 

History

1 Commit

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 

Repository files navigation

Skills

A curated Agent Skills catalog for practicing the AIgile SDLC: a spec-first, test-first, verification-driven workflow for AI-assisted software delivery.

Quick Start

List available skills:

npx skills add jae-labs/skills --list

Install all skills:

npx skills add jae-labs/skills --skill '*'

Install one skill:

npx skills add jae-labs/skills --skill security-and-hardening

Goal And Purpose

This catalog curates and integrates tools and practices for AI-assisted software delivery. Instead of reinventing standard workflows, it references and adapts community skills.

Because agent tools and skill formats evolve quickly, this catalog remains a living integration layer.

AIgile SDLC combines:

  • Agile and XP feedback loops.
  • Spec-driven development for AI coding agents.
  • TDD and systematic debugging.
  • Context engineering for agent output quality.
  • Source-driven development against official documentation.
  • Doubt-driven review of non-trivial decisions.
  • Architecture records and review gates.
  • Structural code quality review.
  • Agent Skills for dynamic, low-token procedural context.
  • Verification before completion.
  • CI/CD automation and staged rollouts.
  • Observability and deprecation discipline.
  • Ongoing maintenance: simplification, hardening, optimization, architecture deepening.
flowchart LR
    I["Ideation"] --> P["Planning"] --> IM["Implementation"] --> V["Verification"] --> R["Review"] --> D["Deploy"] --> M["Maintenance"]
Loading

AIgile SDLC Workflow

flowchart TD
    Start["Idea or request"] --> A["1. global-rules"]
    A --> B["2. context-engineering"]
    B --> C["3. brainstorming"]
    C --> D{"Need deeper<br>challenge?"}
    D -->|yes| E["4. grill-me"]
    D -->|no| F["Spec or issue"]
    E --> F
    F --> G["5. triage"]
    G --> H{"Architecture<br>decision?"}
    H -->|yes| I["ADR / notes"]
    H -->|no| J["6. writing-plans"]
    I --> J
    J --> K["7. using-git-worktrees"]
    K --> L["8. executing-plans"]
    L --> L2{"Non-trivial<br>decision?"}
    L2 -->|yes| L3["11. doubt-driven-development"]
    L2 -->|no| L4{"Framework<br>code?"}
    L3 --> L4
    L4 -->|yes| L5["10. source-driven-development"]
    L4 -->|no| L6{"UI or API?"}
    L5 --> L6
    L6 -->|UI| L7["12. frontend-ui-engineering"]
    L6 -->|API| L8["13. api-and-interface-design"]
    L7 --> M["9. test-driven-development"]
    L8 --> M
    M --> N{"Failure?"}
    N -->|yes| O["14. systematic-debugging"]
    O --> M
    N -->|no| N2{"Browser-facing?"}
    N2 -->|yes| P1["16. browser-testing-with-devtools"]
    N2 -->|no| P["15. verification-before-completion"]
    P1 --> P
    P --> Q["17. thermo-nuclear-code-quality-review"]
    Q --> R["18. requesting-code-review"]
    R --> S["19. receiving-code-review"]
    S --> T["20. md-docs"]
    T --> U["21. create-pr"]
    U --> V["22. finishing-a-development-branch"]
    V --> W["Review, merge, measure"]
    W --> W2{"Deploying?"}
    W2 -->|yes| W3["Deploy skills<br>23-26"]
    W2 -->|no| MT{"Maintenance<br>needed?"}
    W3 --> MT
    MT -->|yes| X["Maintenance skills<br>27-30"]
    MT -->|no| Done["Done"]
Loading

Folder Structure

All installable skills live in one flat namespace:

Folder Contents
skills/<skill-name>/ One directory per installable skill. Each directory contains a SKILL.md and any supporting files for that skill.

The Authors column shows upstream sources in owner/repo:skillname format. Local copies can differ significantly — some skills were rewritten or merged for this repo. Some are wired to route into or hand off to other skills. The flat layout keeps every skill addressable by name.

Skill Catalog

AIgile SDLC Core

# follows the main workflow above.

Category # Skill Purpose Authors
Foundation 1 global-rules Global engineering posture, concise communication, documentation lookup rules, and mandatory skill discovery. Install first. Local
Foundation 2 context-engineering Feed agents the right information at the right time. Use at session start, when switching from design to implementation, or when output quality degrades. addyosmani/agent-skills:context-engineering
Ideation 3 brainstorming Extract intent, ask one question at a time with guesses, generate forced variation approaches, surface assumptions, produce a full spec (objective, architecture, commands, code style, boundaries, success criteria, out of scope), self-review, and get user approval before implementation. obra/superpowers:brainstorming
Ideation 4 grill-me Stress-test a design or plan through relentless one-question-at-a-time interrogation with confidence scoring, decision tree resolution, codebase grounding, and anti-rationalization checks until every branch has a concrete resolution. mattpocock/skills:grill-me
Ideation 5 triage Convert issues into durable agent-ready briefs with acceptance criteria and scope boundaries. mattpocock/skills:triage
Planning 6 writing-plans Convert approved specs into executable plans with dependency graphs, vertical slicing, route/surface mapping, bite-sized TDD steps, task sizing, checkpoints, and risk documentation. Mandatory before multi-step implementation. obra/superpowers:writing-plans, addyosmani/agent-skills:planning-and-task-breakdown
Planning 7 using-git-worktrees Set up or verify isolated workspaces before implementation. obra/superpowers:using-git-worktrees
Implementation 8 executing-plans Execute a written implementation plan task by task with incremental delivery, traceability back to the approved request, review checkpoints, scope discipline, git discipline, and simplicity-first rules. obra/superpowers:executing-plans, addyosmani/agent-skills:incremental-implementation, addyosmani/agent-skills:git-workflow-and-versioning
Implementation 9 test-driven-development Enforce red-green-refactor for features, bug fixes, behavior changes, and refactors. obra/superpowers:test-driven-development
Implementation 10 source-driven-development Ground every framework decision in official documentation. Verify, cite sources, and flag what's unverified. addyosmani/agent-skills:source-driven-development
Implementation 11 doubt-driven-development Subject non-trivial decisions to adversarial fresh-context review. CLAIM → EXTRACT → DOUBT → RECONCILE → STOP. addyosmani/agent-skills:doubt-driven-development
Implementation 12 frontend-ui-engineering Build production-quality UIs with accessibility, design system adherence, responsive design, no AI aesthetic, and no fake preview surfaces standing in for requested product behavior. addyosmani/agent-skills:frontend-ui-engineering
Implementation 13 api-and-interface-design Design stable, well-documented APIs and interfaces that are hard to misuse. Contract-first, Hyrum's Law, consistent errors. addyosmani/agent-skills:api-and-interface-design
Verification 14 systematic-debugging Investigate root cause before fixes. Stop-the-line rule, error-specific triage, git bisect, safe fallbacks, untrusted data treatment. obra/superpowers:systematic-debugging, addyosmani/agent-skills:debugging-and-error-recovery
Verification 15 verification-before-completion Require fresh evidence before claiming work is complete, fixed, or passing. obra/superpowers:verification-before-completion
Verification 16 browser-testing-with-devtools Test and debug in real browsers via Chrome DevTools MCP. Mandatory for browser-facing work before completion claims. Inspect DOM, capture console errors, analyze network requests, profile performance, verify accessibility. addyosmani/agent-skills:browser-testing-with-devtools
Review 17 thermo-nuclear-code-quality-review Run an extremely strict maintainability review focused on structural quality, code judo, file size, and abstraction health. cursor/plugins:thermo-nuclear-code-quality-review
Review 18 requesting-code-review Request an independent code review after implementation and verification. obra/superpowers:requesting-code-review
Review 19 receiving-code-review Address review feedback rigorously and track disposition of each finding. obra/superpowers:receiving-code-review
Deploy 20 md-docs Maintain README.md, AGENTS.md, and CLAUDE.md with a strict human-vs-agent audience split. Routes ADR writing for architectural decisions. PaulRBerg/agent-skills:md-docs, addyosmani/agent-skills:documentation-and-adrs
Deploy 21 create-pr Inspect local changes and create high-quality PR descriptions through the GitHub CLI. Local
Deploy 22 finishing-a-development-branch Decide how to integrate completed branch work after verification. obra/superpowers:finishing-a-development-branch
Deploy 23 ci-cd-and-automation Automate quality gates so no change reaches production without passing tests, lint, type checking, and build. Shift left, faster is safer, pre-commit hygiene, GitHub Actions, deployment strategies, feature flags, staged rollouts, rollback plans. addyosmani/agent-skills:ci-cd-and-automation
Deploy 24 deprecation-and-migration Remove code that no longer earns its keep and move users safely from old to new. Deprecation decision framework, compulsory vs advisory, strangler pattern, adapter pattern, feature flag migration, zombie code identification. addyosmani/agent-skills:deprecation-and-migration
Deploy 25 observability-and-instrumentation Instrument code so production behavior is visible and diagnosable. Structured logging, RED/USE metrics, OpenTelemetry tracing, symptom-based alerting, correlation IDs, cardinality discipline. addyosmani/agent-skills:observability-and-instrumentation
Deploy 26 shipping-and-launch Ship with confidence. Pre-launch checklist (code quality, security, performance, accessibility, infrastructure, documentation), feature flag lifecycle, staged rollout, monitoring, rollback strategy, post-launch verification. addyosmani/agent-skills:shipping-and-launch
Maintenance 27 code-simplification Simplify code for clarity without changing behavior. Chesterton's Fence, language-specific guidance, Rule of 500. addyosmani/agent-skills:code-simplification
Maintenance 28 security-and-hardening Harden code against vulnerabilities. OWASP Top 10, threat modeling, input validation, secrets management, SSRF prevention, LLM security. addyosmani/agent-skills:security-and-hardening
Maintenance 29 performance-optimization Measure-first performance optimization. Core Web Vitals, symptom-based triage, N+1 detection, bundle analysis, caching. addyosmani/agent-skills:performance-optimization
Maintenance 30 improve-codebase-architecture Find deepening opportunities that improve locality, leverage, and testability. mattpocock/skills:improve-codebase-architecture

Supporting Skills

Category Skill Purpose Authors
Communication internal-comms Draft internal updates, FAQs, newsletters, reports, and incident/project communications. anthropics/skills:internal-comms
Communication fuck-slop Remove AI-writing tells from prose when explicitly requested. JuliusBrussee/skills:fuck-slop
Communication frontend-slides Create zero-dependency, animation-rich 16:9 HTML presentations from scratch or PPTX imports. zarazhangrui/frontend-slides
Review principal-sre-reviewer Explicit-only Principal SRE pull request review focused on reliability, operability, observability, deployability, security, CI/CD, dependency/version safety, and risk. Drafts findings and inline comments without writing to GitHub. Local
Foundation mermaid-diagrams Create or edit Mermaid/software diagrams when explicitly requested. softaworks/agent-toolkit:mermaid-diagrams
Foundation excalidraw-diagram-generator Generate Excalidraw flowcharts, relationship maps, and system design diagrams in JSON format. github/awesome-copilot:excalidraw-diagram-generator
Token Control cocoindex-code Semantic code search and indexing over the codebase. cocoindex-io/cocoindex-code:cocoindex-code
Token Control caveman Explicit-only ultra-compressed communication mode. JuliusBrussee/caveman:caveman
Token Control handoff Compact a session into a handoff document for another agent or future session. mattpocock/skills:handoff
Skill Authoring skill-creator Create, improve, evaluate, benchmark, and package Agent Skills. anthropics/skills:skill-creator
Learning teach Build structured learning material and learning records for a topic. mattpocock/skills:teach
Terraform terraform-style-guide Generate or review Terraform HCL using HashiCorp style conventions. hashicorp/agent-skills:terraform-style-guide
Terraform refactor-module Refactor Terraform modules or split monolithic Terraform configurations into reusable modules. hashicorp/agent-skills:refactor-module
SRE postmortem-writer Conduct blameless incident writeups (RCAs) and generate structured postmortem reports. Local
System Design system-design-specialist Emulate a Principal Engineer using Non-Abstract Large-Scale System Design (NALSD) and progressive refinement. Local

Reference Material

These sources align with AIgile SDLC:

Compatibility

These skills follow the Agent Skills SKILL.md convention. The Skills CLI can install them into agents like Claude Code, Cursor, GitHub Copilot, Cline, and Gemini CLI.

Map platform-specific tools mentioned in any skill to their closest local equivalents.

Contributing

See CONTRIBUTING.md.

Agent Notes

See AGENTS.md.

License

See LICENSE.

Stars

Star History

Star History Chart

Releases

Packages

Used by

Contributors

Languages