ai-agent-skills
AGENTS.md216 starsChanged 6 months agoMITView on GitHub
This repository contains AI agent skills — structured markdown skill files for production-grade AI-assisted development. Before a skill is merged: - [ ] Applied to at least one real agent session - [ ] All sections filled (no "TODO" or empty sections) - [ ] Verification criteria are concrete (not "make sure it works") - [ ] Anti-rationalization table has at least 3 entries - [ ] Added to skills/index.json
We split this AGENTS.md into the rules it gives an agent
8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Adding skills: Follow the template in docs/skill-template.md exactly.
- Distinctive: line 9
Editing skills: Only edit the skill you're asked to change.
- Distinctive: line 10
Testing skills: Test by applying the skill to a real agent session and verifying the output quality.
- Distinctive: line 28
Every step has a verification gate — "seems right" is never sufficient
- Very specific: line 35
Create SKILL.md using the template from docs/skill-template.md
- Very specific: line 36
Add to skills/index.json for the GitHub Pages search to pick it up
- Distinctive: line 37
Update README.md skills table
- Distinctive: line 41
Before a skill is merged:
≈ “Common” is an estimate: at least 1 in 5 of the 36,368 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
AI Agent Instructions for This Repository
This repository contains AI agent skills — structured markdown skill files for production-grade AI-assisted development.
When Working in This Repository
- Reading skills: Skills are in
skills/<name>/SKILL.md. Read the full SKILL.md before making any edits. - Adding skills: Follow the template in
docs/skill-template.mdexactly. Every section is required. - Editing skills: Only edit the skill you're asked to change. Don't refactor adjacent skills.
- Testing skills: Test by applying the skill to a real agent session and verifying the output quality.
Repository Structure
skills/ ← All 30 skill files (SKILL.md per directory)
agents/ ← Specialist agent persona definitions
references/ ← Supporting checklists (security, performance, etc.)
docs/ ← Setup guides, skill anatomy, template
.claude/ ← Claude Code slash commands
.gemini/ ← Gemini CLI commands
.github/ ← GitHub Actions workflows
docs/ ← GitHub Pages site (index.html)Key Principles
- Skills are workflows, not prose — agents follow steps, not skim docs
- Every step has a verification gate — "seems right" is never sufficient
- Skills include anti-rationalization tables — common excuses + rebuttals
- Skills are minimal — only what's needed to guide the agent through the task
When Adding a New Skill
- Create directory
skills/<kebab-name>/ - Create
SKILL.mdusing the template fromdocs/skill-template.md - Add to
skills/index.jsonfor the GitHub Pages search to pick it up - Update
README.mdskills table
Validation
Before a skill is merged:
- Applied to at least one real agent session
- All sections filled (no "TODO" or empty sections)
- Verification criteria are concrete (not "make sure it works")
- Anti-rationalization table has at least 3 entries
- Added to
skills/index.json
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 28,693 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 repository34 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- agent-qa-run-reviewSkill
- ai-output-validationSkill
- api-designSkill
- ci-cd-pipelinesSkill
- citruscoastaiSkill
- code-explanationSkill
- code-reviewSkill
- context-loadingSkill
- debugging-methodologySkill
- documentationSkill
- error-handlingSkill
- frontend-engineeringSkill
- git-workflowSkill
- goal-driven-executionSkill
- hallucination-preventionSkill
- idea-to-specSkill
- incremental-codingSkill
- integration-testingSkill
- meeting-notes-to-tasksSkill
- multi-agent-orchestrationSkill
- observabilitySkill
- performance-optimizationSkill
- production-deploymentSkill
- prompt-injection-defenseSkill
- rag-and-memorySkill
- refactoringSkill
- research-and-summarizeSkill
- security-hardeningSkill
- simplicity-firstSkill
- surgical-changesSkill
- task-decompositionSkill
- test-driven-developmentSkill
- think-before-codingSkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
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.