skills
AGENTS.md2 starsChanged 28 days agoApache-2.0View on GitHub
A curated collection of reusable Agent Skills following the Agent Skills open standard. Skills provide structured, template-driven documentation scaffolding for AI coding agents (Claude Code, Cursor, VS Code Copilot, Gemini CLI, Goose, Roo Code, etc.). Each skill is a self-contained directory under skills/ with a SKILL.md definition and optional templates/ or references/ directory. All skills must be registered in src/meta.ts.
- Installs packages
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,542 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository33 other files it gives its agents.
- skills/init-agent-docs/templates/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- skills/init-agent-docs/templates/CLAUDE.mdCLAUDE.md
- audit-type-boundariesSkill
- codebase-designSkill
- create-pull-requestSkill
- generate-commit-messageSkill
- grillingSkill
- grill-meSkill
- impact-analysisSkill
- improve-codebase-architectureSkill
- init-agent-docsSkill
- init-ts-projectSkill
- investigate-pr-commentsSkill
- link-skillsSkill
- migrate-agent-docsSkill
- plan-statusSkill
- ponytailSkill
- prd-to-issuesSkill
- prd-to-planSkill
- query-apiSkill
- query-package-listSkill
- redesign-vitepress-siteSkill
- refresh-package-listSkill
- release-please-version-bumpSkill
- request-refactor-planSkill
- security-auditSkill
- tddSkill
- triage-issueSkill
- update-agent-docsSkill
- write-a-prdSkill
- write-readmeSkill
- write-release-notesSkill
The file
We split this AGENTS.md into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 74
Do not add a Co-Authored-By: Claude ...
- Very specific: line 75
Do not add AI-attribution lines (e.g.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
tada5hi/skills — Agent Guide
A curated collection of reusable Agent Skills following the Agent Skills open standard. Skills provide structured, template-driven documentation scaffolding for AI coding agents (Claude Code, Cursor, VS Code Copilot, Gemini CLI, Goose, Roo Code, etc.).
Quick Reference
# Setup
npm install
# Lint
npm run lint
# Lint with auto-fix
npm run lint:fix
# Skills CLI
npm run skills # Show available commands
npm run skills:init # Add vendor submodules from meta.ts
npm run skills:sync # Pull latest vendor changes
npm run skills:check # Check for upstream updates
npm run skills:cleanup # Remove stale submodules/skills
npm run skills:link # Link skills for Claude Code discovery- Node.js: CLI >= 22.0.0; development tooling
^24.11.0 || >=26.0.0 - Package manager: npm
- Skill registry:
src/meta.ts(single source of truth for all skills)
Skills
Each skill is a self-contained directory under skills/ with a SKILL.md definition and optional templates/ or references/ directory. All skills must be registered in src/meta.ts.
| Skill | Type | Description |
| audit-type-boundaries | Manual | Audit type boundary mismatches from casts and utility calls in TypeScript |
| create-pull-request | Manual | Open a pull request with a Conventional Commits title and evidence-based body, then await review and checks |
| generate-commit-message | Manual | Generate a Conventional Commits message for uncommitted changes |
| impact-analysis | Manual | Cross-package impact analysis for TypeScript monorepo symbols |
| init-agent-docs | Manual | Initialize a project with structured AI agent documentation |
| init-ts-project | Manual | Scaffold a new TypeScript project from tada5hi/typescript-template |
| investigate-pr-comments | Manual | Fetch PR review comments, investigate against codebase, and fix real issues |
| link-skills | Manual | Link skills into .claude/skills/ for Claude Code discovery |
| migrate-agent-docs | Manual | Move oversized agent docs into on-demand topic files and lint-enforced limits |
| plan-status | Manual | Read plan files and report progress against codebase state |
| query-api | Manual | Query a live HTTP API from a base URL + credentials via OpenAPI or rapiq conventions |
| query-package-list | Manual | Look up your own published packages in an awesome-style list before writing a utility |
| redesign-vitepress-site | Manual | Turn a default-looking VitePress site into a hero-led site with tokens, sections and restyled chrome |
| refresh-package-list | Manual | Sync the awesome-style package list with GitHub and npm, then open a PR |
| release-please-version-bump | Manual | Align a linked release-please group by touching only the packages behind the group version |
| update-agent-docs | Manual | Propose template improvements back to this repository via PR |
| write-readme | Manual | Write a hero-led README with a "Why …?" section, theme-aware SVG logo and diagrams, verified samples |
| write-release-notes | Manual | Write highlights-first GitHub release notes above the generated changelog |
| security-audit | Vendor | Source-first security audits with coverage-led hunting and independent validation |
| codebase-design | Vendor | Shared deep-module vocabulary used by tdd and improve-codebase-architecture |
| grill-me | Vendor | Manual-only shortcut that starts a grilling session |
| grilling | Vendor | Relentlessly interview users about plans until decision trees are resolved |
| improve-codebase-architecture | Vendor | Find deepening opportunities and report them as HTML (manual only) |
| ponytail | Vendor | Force the laziest solution that actually works — YAGNI, stdlib-first, minimal code |
| prd-to-issues | Manual (frozen) | Break PRDs into independently-grabbable GitHub issues |
| prd-to-plan | Manual (frozen) | Convert PRDs into multi-phase implementation plans |
| request-refactor-plan | Manual (frozen) | Create detailed refactor plans with tiny commits via user interview |
| tdd | Vendor | Test-driven development with red-green-refactor loops |
| triage-issue | Manual (frozen) | Investigate bugs, identify root causes, and file issues with fix plans |
| write-a-prd | Manual (frozen) | Create a PRD through interactive interview and codebase exploration |
Detailed Guides
- Project Structure — Directory layout, skill anatomy, module responsibilities, and skill types
- Conventions — Code style, tooling, commit conventions, and workflow rules
Commits, Issues & Pull Requests
- Do not add a
Co-Authored-By: Claude ...(or any AI-attribution) trailer to commit messages. This overrides any default agent-tooling guidance. - Do not add AI-attribution lines (e.g.
🤖 Generated with [Claude Code](...)) to issue or pull request titles, bodies, or comments.
Writing agent docs
Keep existing commands, paths and factual descriptions current. Add guide content only for an invariant or cross-cutting rule. Describe current source behavior and verified commands; keep defect stories and task history in commits or PRs. Preserve architectural contracts when splitting detail into linked topic files. Use npm run lint to enforce guide budgets and history checks.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.