claude-code-skills / skills
AGENTS.md1 starsChanged 46 days agoMITView on GitHub
The published Claude Code skills. Each skill is a kebab-case directory containing a SKILL.md entry point (YAML frontmatter + instructions) and, optionally, references/ for long material loaded on demand. Users install by copying a skill directory to ~/.claude/skills/ or uploading it to Claude.ai.
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 repository16 other files it gives its agents.
- agents/AGENTS.mdAGENTS.md
- AGENTS.mdAGENTS.md
- examples/AGENTS.mdAGENTS.md
- .github/AGENTS.mdAGENTS.md
- .github/workflows/AGENTS.mdAGENTS.md
- scripts/AGENTS.mdAGENTS.md
- skills/figma-handoff/AGENTS.mdAGENTS.md
- skills/github-pr-review/AGENTS.mdAGENTS.md
- skills/gmail-workflow/AGENTS.mdAGENTS.md
- skills/skill-compliance-checklist/AGENTS.mdAGENTS.md
- skills/skill-compliance-checklist/references/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- figma-handoffSkill
- github-pr-reviewSkill
- gmail-workflowSkill
- skill-compliance-checklistSkill
The file
We split this AGENTS.md into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
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 27
If a SKILL.md nears 5,000 words (largest today: figma-handoff, ~1,100), move detail into references/.
- Distinctive: line 33
Run the local CI command from the root AGENTS.md.
- Very specific: line 35
No skill has eval/, test/, or tests/.
≈ “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
<!-- Parent: ../AGENTS.md --><!-- Generated: 2026-06-30 | Updated: 2026-09-27 -->skills/
Purpose
The published Claude Code skills. Each skill is a kebab-case directory containing a SKILL.md entry point (YAML frontmatter + instructions) and, optionally, references/ for long material loaded on demand. Users install by copying a skill directory to ~/.claude/skills/ or uploading it to Claude.ai.
Subdirectories
| Directory | Purpose |
skill-compliance-checklist/ | Meta-skill auditing skills against Anthropic's Jan 2026 guide — the repo's pre-publish gate (see skill-compliance-checklist/AGENTS.md) |
figma-handoff/ | Figma-to-code handoff via the Figma MCP server (see figma-handoff/AGENTS.md) |
github-pr-review/ | Six-step PR review via the gh CLI (see github-pr-review/AGENTS.md) |
gmail-workflow/ | Draft, triage, search, and draft management via the Gmail MCP server (see gmail-workflow/AGENTS.md) |
For AI Agents
Working In This Directory
- Directory names are kebab-case and each must contain a file named exactly
SKILL.md(scripts/attention-check.shscores a skill directory without one at 75). - Frontmatter: CI requires
name(kebab-case) and a single-linedescription(≤ 1024 chars, no</>). Repo convention addslicense: MITandmetadata(author,version, optionalmcp-server/based-on).user-invocableis set explicitly only on the checklist (true). - Descriptions follow WHAT + WHEN + negative triggers ("Do NOT use for…").
- Bump
metadata.versionon every functional change (all at 1.0.0 except the checklist at 1.1.0). - If a SKILL.md nears 5,000 words (largest today:
figma-handoff, ~1,100), move detail intoreferences/. - Audit with explicit paths:
/skill-compliance-checklist skills/{name}/SKILL.md—--alldoes not scan this repo. - New or renamed skills also need the README Skills table updated.
Testing Requirements
- Run the local CI command from the root
AGENTS.md; it must end with "All skills and agents passed validation." - Include
## Error Handling(or## Troubleshooting) and## Examples— CI only warns when missing, but the compliance checklist fails them. - No skill has
eval/,test/, ortests/; adding one clears that skill's attention-check "no evals" signal.
Common Patterns
- The three companion skills share one layout: overview →
## Available (MCP) Tools→## Instructions(numbered steps) →## Error Handlingtable →## Examples(three each) →## Important. The checklist uses Instructions → Troubleshooting → Examples. - MCP-backed skills declare
metadata.mcp-server(figma,gmail) and list the exact tool names they call. gmail-workflowandgithub-pr-reviewrequire explicit user confirmation before anything is sent or submitted — keep that guard.
Dependencies
Internal
skill-compliance-checklist/— validates the other skills andexamples/*/SKILL.mdgithub-pr-review/overlapsagents/code-reviewer.md(different severity scale)
External
- Figma MCP server (
figma-handoff), Gmail MCP server (gmail-workflow),ghCLI (github-pr-review)
<!-- MANUAL: -->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.