jara-r-kjara-r-k/claude-code-skills/AGENTS.md

claude-code-skills

AGENTS.md1 starsChanged 46 days agoMITView on GitHub

Public (MIT) repository of reusable Claude Code skills, agents, and example templates. Published content is Markdown only: skills are SKILL.md files Claude Code loads by trigger phrase, agents are YAML-fronted .md subagent definitions, and examples are copy-and-customise templates. There is no build step and no package manifest. The only executable code is the CI validator (inline Bash in .github/workflows/validate-skills.yml) and scripts/attention-check.sh, a health scanner consumed by the ~/projects Attention Hub. - No local test runner. CI (.github/workflows/validate-skills.yml) runs on push/PR…

Reads credentials

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
- 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

7 rules. About 0 are common, 1 is distinctive, and 6 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.
6
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Keep published content Markdown-only — no package.json, Makefile, or build output.
line 35
Very specific:
Follow CLAUDE.md for frontmatter and section rules.
line 36
Distinctive:
Keep the README tables in sync when adding, renaming, or removing a skill, agent, or example.
line 39
Very specific:
Never write to ~/projects/raw/ (human-curated wiki layer, protected by a PreToolUse hook).
line 41
Very specific:
No local test runner. CI (.github/workflows/validate-skills.yml) runs on push/PR touching skills/, agents/, examples/, or …
line 45
Very specific:
Run the identical checks locally from the repo root: awk '/^ run: \|/{f=1;next} f' .github/workflows/validate-skills.yml | sed 's/^ //' | …
line 46
Very specific:
Keep description: on one line — CI reads only that line, so description: > fails and description: | escapes the length check.
line 57

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

AGENTS.md

claude-code-skills

Purpose

Public (MIT) repository of reusable Claude Code skills, agents, and example templates. Published content is Markdown only: skills are SKILL.md files Claude Code loads by trigger phrase, agents are YAML-fronted .md subagent definitions, and examples are copy-and-customise templates. There is no build step and no package manifest. The only executable code is the CI validator (inline Bash in .github/workflows/validate-skills.yml) and scripts/attention-check.sh, a health scanner consumed by the ~/projects Attention Hub.

Key Files

File	Description
`CLAUDE.md`	Canonical conventions — skill/agent frontmatter format, success criteria, gotchas. Read before editing any skill or agent
`README.md`	Public docs — skills/agents/examples tables, install commands, MCP requirements, CI summary
`LICENSE`	MIT licence
`.gitignore`	Ignores `.omc/`, `.claude/`, `.DS_Store`, and secret-bearing patterns (`.env`, `.pem`, `*.key`, `credentials.json`, …)
`.claudeignore`	Claude Code context exclusions (generic Node/build/log/env patterns; header comment mislabels the repo as TypeScript/Node)

Subdirectories

Directory	Purpose
`skills/`	Published skills, one kebab-case directory each (see `skills/AGENTS.md`)
`agents/`	Subagent definition files (see `agents/AGENTS.md`)
`examples/`	Adaptation templates with placeholders (see `examples/AGENTS.md`)
`scripts/`	`attention-check.sh` — Attention Hub scanner, not used by CI (see `scripts/AGENTS.md`)
`.github/`	GitHub Actions CI (see `.github/AGENTS.md`)
`.claude/`	Optional gitignored local configuration; contents vary by checkout.
`.omc/`	Gitignored oh-my-claudecode session state. Generated — do not edit or document

For AI Agents

Working In This Directory

Keep published content Markdown-only — no package.json, Makefile, or build output.
Follow CLAUDE.md for frontmatter and section rules; bump metadata.version on every functional skill change.
Conventional commits (feat:, fix:, docs:, chore:); Australian English (colour, behaviour, organise, licence as noun).
/skill-compliance-checklist only audits what it is pointed at: its --all scan covers ~/.claude/skills/, ~/.claude/scheduled-tasks/, and .claude/agents/ — not this repo's skills/. Pass explicit paths (e.g. skills/gmail-workflow/SKILL.md).
Keep the README tables in sync when adding, renaming, or removing a skill, agent, or example.
scripts/attention-check.sh is invoked by path from outside this repo — do not move or rename it.
Never write to ~/projects/raw/ (human-curated wiki layer, protected by a PreToolUse hook).

Testing Requirements

No local test runner. CI (.github/workflows/validate-skills.yml) runs on push/PR touching skills/, agents/, examples/, or .github/workflows/.
Run the identical checks locally from the repo root: awk '/^ run: \|/{f=1;next} f' .github/workflows/validate-skills.yml | sed 's/^ //' | bash
Hard failures: no --- on line 1, missing or non-kebab name:, missing description:, description > 1024 chars or containing </>, file > 5,000 words (whole file incl. frontmatter). Missing ## Error…/## Troubleshoot… and ## Example… headings only warn.
Agents are checked only for the --- delimiter and a non-empty name:; references/ files are never validated.
scripts/: bash scripts/attention-check.sh 2>/dev/null | python3 -m json.tool.

Common Patterns

Skill: skills/{kebab-name}/SKILL.md, optional references/ for long material (progressive disclosure).
Agent: agents/{kebab-name}.md with name, description, license, model: sonnet, tools.
Example: examples/{name}/SKILL.md + references/ customisation guide; placeholders stay generic.
Keep description: on one line — CI reads only that line, so description: > fails and description: | escapes the length check.

Dependencies

Internal

skills/skill-compliance-checklist/ — manual pre-publish validator for every other skill and example

External

GitHub Actions (ubuntu-latest, actions/checkout@v4) — CI
gh CLI — github-pr-review skill only (CI does not use it)
Figma MCP server (figma-handoff), Gmail MCP server (gmail-workflow)
~/projects/scripts/attention-collect.sh (Attention Hub) — consumes scripts/attention-check.sh

Download now (ZIP)

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

What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free