claude-code-skills / agents
AGENTS.md1 starsChanged 46 days agoMITView on GitHub
Reusable Claude Code subagent definitions. Each .md file has YAML frontmatter (name, description, license, model, tools) and a methodology body. Agents execute focused work when spawned; skills (../skills/) guide behaviour when triggered. Users install by copying into ~/.claude/agents/ or {project}/.claude/agents/.
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.mdAGENTS.md
- examples/AGENTS.mdAGENTS.md
- .github/AGENTS.mdAGENTS.md
- .github/workflows/AGENTS.mdAGENTS.md
- scripts/AGENTS.mdAGENTS.md
- skills/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
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 24
No word limit is enforced for agents.
- Distinctive: line 26
Never add deployment or destructive commands to an agent body.
- Distinctive: line 31
Run the local CI command from the root AGENTS.md.
- Distinctive: line 32
No behavioural tests: spawn the agent in a real context and check it follows its steps and output format.
≈ “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 -->agents/
Purpose
Reusable Claude Code subagent definitions. Each .md file has YAML frontmatter (name, description, license, model, tools) and a methodology body. Agents execute focused work when spawned; skills (../skills/) guide behaviour when triggered. Users install by copying into ~/.claude/agents/ or {project}/.claude/agents/.
Key Files
| File | Description |
code-reviewer.md | Reviews diffs for security (OWASP Top 10), quality, performance, maintainability; findings graded Critical/High/Medium/Low/Nit plus Positives and Suggestions. Tools: Read, Glob, Grep, Bash |
project-setup.md | Explores a new repo, detects the stack from marker files, writes a concise CLAUDE.md (< 80 lines) and a .claude/agents/ + .claude/commands/ skeleton. Tools: Read, Write, Edit, Bash, Glob, Grep |
python-test-runner.md | Activates a venv, runs pytest at the requested scope, reports counts, failure root causes, environment issues, flaky tests, coverage. Tools: Read, Glob, Grep, Bash |
For AI Agents
Working In This Directory
- CI requires only
---on line 1 and a non-emptyname:. Match the existing files anyway:description(with "Do NOT use for…"),license: MIT,model: sonnet, minimaltoolslist. - CI skips
AGENTS.mdby exact basename (it has no frontmatter). Do not rename it or add any other non-agent.mdhere — CI would fail it. - No word limit is enforced for agents; self-enforce concision (current files are 640–710 words).
- "Read-only" in
code-reviewerandpython-test-runneris an instruction, not a tool restriction — both haveBash. Keep the prohibitions in the body. - Never add deployment or destructive commands to an agent body.
- New agents go in the README Agents table (the README install snippet currently copies only
code-reviewerandproject-setup).
Testing Requirements
- Run the local CI command from the root
AGENTS.md. - No behavioural tests: spawn the agent in a real context and check it follows its steps and output format.
Common Patterns
- Numbered steps, then
## Rules;code-reviewerandpython-test-runneralso define an## Output Formatblock. code-reviewer.mdoverlapsskills/github-pr-review/but uses a different severity scale (Critical/High/Medium/Low/Nit vs Critical/Suggestion/Nit) and omits empty severity sections. Align both deliberately if changing either.project-setup.mdmust read an existingCLAUDE.mdbefore touching it and stay idempotent. It must change configuration only, never modify source code, install packages, or run deployment commands.
Dependencies
Internal
skills/github-pr-review/— sibling review methodology (see overlap note)
External
- Git —
code-reviewer(git diff,git log) - Python + pytest; optional pytest-cov, pytest-xdist, pytest-timeout —
python-test-runner
<!-- 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.