agentleFS
Sign inStart free
pahlevikunpahlevikun/levistack/AGENTS.md

levistack

AGENTS.md0 starsChanged 18 days agoMITView on GitHub

One source tree of skills, rules, subagents, commands, and hooks for Claude Code, Cursor, Codex, and the skills CLI. CLAUDE.md is a symlink to this file. Human install steps and the skill-group table live in README.md. Maintaining this repository (anything under the catalog tree or under .cursor/): Using the published catalog (explore, plan, implement, review in this or another repo): Do not invent a parallel workflow when a skill, agent, or rule already covers the work. Claude Code reads .claude/agents,…

The file

We split this AGENTS.md into the rules it gives an agent

5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    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
  1. Very specific:

    Read and follow .cursor/skills/manage-stack/SKILL.md.

    line 11
  2. Very specific:

    Prefer the repo-maintainer agent (.cursor/agents/repo-maintainer.md) for a bounded catalog edit.

    line 12
  3. Distinctive:

    Do not invent a parallel workflow when a skill, agent, or rule already covers the work.

    line 20
  4. Distinctive:

    Do not commit or push unless asked.

    line 63
  5. Very specific:

    No employer-owned content, secrets, or personal paths.

    line 71

≈ “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

levistack

One source tree of skills, rules, subagents, commands, and hooks for Claude Code, Cursor, Codex, and the skills CLI. CLAUDE.md is a symlink to this file.

Human install steps and the skill-group table live in README.md.

Before you change anything

Maintaining this repository (anything under the catalog tree or under .cursor/):

  1. Read and follow .cursor/skills/manage-stack/SKILL.md.
  2. Prefer the repo-maintainer agent (.cursor/agents/repo-maintainer.md) for a bounded catalog edit.

Using the published catalog (explore, plan, implement, review in this or another repo):

  • Load the skill whose description matches the task (skills/<group>/<skill>/SKILL.md).
  • Delegate to a subagent when its description fits; roster and flow: docs/agents/README.md.
  • Apply rules from rules/ (Cursor: generated/cursor/rules/ after sync). Project sessions in this repo also load .cursor/rules/ (for example manage-stack.mdc).

Do not invent a parallel workflow when a skill, agent, or rule already covers the work.

Two write surfaces

SurfaceWhat it isYou edit
CatalogShipped by the plugins and npx skills addskills/, agents/, rules/, commands/, hooks/
Project agent treeHow agents work in this repo only.cursor/agents/, .cursor/rules/, .cursor/skills/

Claude Code reads .claude/agents, .claude/rules, and .claude/skills as directory symlinks to .cursor/. npm run sync recreates them. Never add stack tooling under .claude/; change .cursor/ instead.

Catalog rules are authored as rules/<topic>/*.md. Sync writes Cursor .mdc files to generated/cursor/rules/ (see .cursor-plugin/plugin.json). Only alwaysApply: true files in rules/core/ feed the session-start hook (hooks/README.md).

Commands

Requires Node ≥ 20.

npm run sync          # manifests, README group table, hooks.json, .claude → .cursor symlinks
npm run check         # sync --check, validate, tests
npm run validate
npm test
npm run check:publish # strict; run before making the repo public

After editing a catalog skill or agent:

node skills/agent-authoring/create-agent/scripts/lint.mjs skills/<group>/<name>
node skills/agent-authoring/create-agent/scripts/lint.mjs agents/<name>.md

Local dev (optional): ./scripts/link-skills.sh symlinks catalog skills into ~/.claude/skills and ~/.agents/skills. Statusline: node scripts/install-statusline.mjs (see statusline/README.md).

Verify before reporting done

npm run sync && git diff --exit-code   # fail = generated output changed; stage it with your source edit
npm run validate
npm test
claude plugin validate . --strict      # when plugin manifests or the tree changed

Do not commit or push unless asked.

Boundaries

  • Edit only catalog paths above, plus docs/agents/README.md when the agent roster changes, upstream.json when an upstream skill is added or removed, and .cursor/ for stack tooling.
  • Upstream skills (upstream.json) live in their own repos. Sync lists them in the marketplace as GitHub sources and as dependencies of the root plugin. Edit them in their own repo, never copy them into skills/.
  • Never hand-edit .claude-plugin/, .cursor-plugin/, .codex-plugin/, .antigravity-plugin/, the root plugin.json, skills/<group>/.claude-plugin/, generated/, hooks/hooks.json, or the generated tables in README.md and hooks/README.md. Run npm run sync.
  • Skill folder name = frontmatter name. description is one quoted line, ≤ 1024 characters, and says when to use the skill.
  • No employer-owned content, secrets, or personal paths. No repo-root THIRD_PARTY.md; keep import licenses and UPSTREAM.md beside copied code.

Where to look

I want to…Look at…
Add or change a shipped skillskills/<group>/<name>/SKILL.md plus that group's DESCRIPTION.md; follow manage-stack
Add or change a catalog subagentagents/<name>.md, docs/agents/README.md
Add or change a rulerules/<topic>/<name>.md
Add a slash commandcommands/<name>.md
Add or change a hookhooks/<name>.mjs, hooks/lib/, hooks/README.md
Change how this repo is maintained.cursor/skills/manage-stack/, .cursor/rules/manage-stack.mdc
Understand sync / validatescripts/sync.mjs, scripts/validate.mjs
Install or group listREADME.md

Runtime entry points

RuntimeReads
Skills CLIskills/**/SKILL.md via npx skills add
Claude Code plugin.claude-plugin/, per-group skills/<group>/.claude-plugin/
Claude Code (this repo).claude/{agents,rules,skills} → .cursor/, this file via CLAUDE.md
Cursor plugin.cursor-plugin/ → every skill path listed, agents/, generated/cursor/rules/; or scripts/link-skills.sh --cursor
Cursor (this repo).cursor/{agents,rules,skills}/ (source)
Codexthe Claude marketplace in .claude-plugin/marketplace.json, then .codex-plugin/plugin.json, this file
Antigravityroot plugin.json (only name and description are allowed) and .antigravity-plugin/plugin.json, both generated, or scripts/link-skills.sh --antigravity
Traeno plugin format: scripts/link-skills.sh --trae into ~/.trae/{skills,agents,rules}

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