agent-skills
AGENTS.md3 starsChanged 7 days agoMITView on GitHub
Every skill must have a SKILL.md with YAML frontmatter: The body of SKILL.md contains the full instructions the agent follows when the skill is invoked. This repository follows a bilingual (English/Japanese) pattern: SKILL.md body must be written in English. Use a Language Rules section to define auto-detection and reference file selection for Japanese support. Skills should auto-detect the user's language and reference the appropriate files. For detailed SKILL.md authoring conventions, see docs/skill-style-guide.md. Before submitting, verify: 実装時のコーディングルールは以下のファイルに従ってください: - docs/coding-rules.md — spec-rules-init…
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 31,774 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 repository24 other files it gives its agents.
- agent-delegateSkill
- cmux-delegateSkill
- cmux-forkSkill
- cmux-second-opinionSkill
- handoverSkill
- harness-initSkill
- harness-loopSkill
- harness-planSkill
- ja-humanizerSkill
- magiSkill
- mcp-convertSkill
- skill-suggestSkill
- spec-codeSkill
- spec-evaluateSkill
- spec-generatorSkill
- spec-implementSkill
- spec-inspectSkill
- spec-orchestrateSkill
- spec-reviewSkill
- spec-rules-initSkill
- spec-testSkill
- spec-to-issueSkill
- spec-workflow-initSkill
- spec-writingSkill
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
16 rules. About 0 are common, 7 are distinctive, and 9 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
9
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 21
Create a directory under skills/ with your skill name (kebab-case)
- Very specific: line 22
Add a SKILL.md with the required frontmatter
- Very specific: line 23
Add any reference files under references/
- Very specific: line 24
Update the skills table in README.md and README.ja.md
- Very specific: line 28
Every skill must have a SKILL.md with YAML frontmatter:
- Distinctive: line 44
name: Must match the directory name exactly
- Distinctive: line 45
description: Keep under 1024 characters.
- Very specific: line 46
license: Use MIT unless you have a specific reason not to
- Very specific: line 55
SKILL.md body must be written in English: Use a Language Rules section to define auto-detection and reference file selection for Japanese …
- Distinctive: line 61
Write clear, concise instructions
- Distinctive: line 62
Avoid hardcoding tool names or MCP server references — use generic descriptions instead
- Very specific: line 63
Keep SKILL.md under 500 lines
- Distinctive: line 64
Use AskUserQuestion for interactive decisions, not free-form text prompts
- Distinctive: line 73
All referenced files exist
- Very specific: line 74
No hardcoded MCP tool names (e.g., mcp__serena__, Context7)
- Distinctive: line 97
Verify the handover against the current repository state before trusting it:
≈ “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
Contributing to agent-skills
Repository Structure
agent-skills/
├── skills/
│ ├── skill-name/
│ │ ├── SKILL.md # Skill definition (required)
│ │ └── references/ # Supporting documents (optional)
│ │ ├── guide.md # English version
│ │ └── guide.ja.md # Japanese version
│ └── ...
├── LICENSE
├── README.md
└── AGENTS.mdAdding a New Skill
- Create a directory under
skills/with your skill name (kebab-case) - Add a
SKILL.mdwith the required frontmatter - Add any reference files under
references/ - Update the skills table in
README.mdandREADME.ja.md
SKILL.md Format
Every skill must have a SKILL.md with YAML frontmatter:
---
name: skill-name # Must match directory name
description: |
Brief description of what the skill does.
Include trigger phrases for discoverability.
license: MIT
---The body of SKILL.md contains the full instructions the agent follows when the skill is invoked.
Guidelines
- name: Must match the directory name exactly
- description: Keep under 1024 characters. Include trigger phrases in all supported languages for discoverability
- license: Use
MITunless you have a specific reason not to
Bilingual Support
This repository follows a bilingual (English/Japanese) pattern:
*.md— English version (primary)*.ja.md— Japanese version
SKILL.md body must be written in English. Use a Language Rules section to define auto-detection and reference file selection for Japanese support.
Skills should auto-detect the user's language and reference the appropriate files.
Code Style
- Write clear, concise instructions
- Avoid hardcoding tool names or MCP server references — use generic descriptions instead
- Keep SKILL.md under 500 lines
- Use AskUserQuestion for interactive decisions, not free-form text prompts
For detailed SKILL.md authoring conventions, see docs/skill-style-guide.md.
Testing
Before submitting, verify:
SKILL.mdfrontmatternamematches directory name- All referenced files exist
- No hardcoded MCP tool names (e.g.,
mcp__serena__,Context7) - SKILL.md is under 500 lines
bash scripts/check-skill-line-budget.shpasses for the whole repositorybash skills/agent-delegate/references/scripts/tests/check_skill_quality.sh skills/*/SKILL.mdpasses for the whole repository- Changes to the quality checker pass
bash skills/agent-delegate/references/scripts/tests/check_skill_quality_contract.sh - For reference-split changes,
bash scripts/tests/check-skill-reference-split-contract.shpasses
コーディングルール
実装時のコーディングルールは以下のファイルに従ってください:
- docs/coding-rules.md — spec-rules-init で生成された品質ルール集
- docs/review_rules.md — コードレビュー・セカンドオピニオン用のレビュー基準
Development Workflow
開発フロー(Issue → 実装 → PR)は以下のファイルに従ってください:
- docs/issue-to-pr-workflow.md — spec-workflow-init で生成された開発ワークフロー
<!-- handover:start -->Session Handover
At session start, if local handover files exist (handover.md or .handover/current.md), read them before making changes.
Verify the handover against the current repository state before trusting it:
- current branch
- current HEAD
- working tree status
- referenced important files
Summarize inherited context, note stale or conflicting information, and continue from Next Action only when it is safe and obvious. Ask before destructive, external-facing, or ambiguous actions.
<!-- handover:end -->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.

