agentleFS
Sign inStart free
jamesm-devjamesm-dev/tateru.dev/AGENTS.md

tateru.dev

AGENTS.md1 starsChanged 13 days agoMITView on GitHub

For AI coding tools other than Claude Code — Codex, Cursor, Gemini CLI, Copilot, and anything else that reads AGENTS.md. tateru is Claude-first. It is built for Claude Code, and the complete instructions live in CLAUDE.md. This file does not repeat them. It tells you what carries over to your tool, what does not, and how to follow the workflow by hand. 1. CLAUDE.md — the project rules. They apply to you too, except where they name a Claude Code…

  • 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 repository9 other files it gives its agents.

The file

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

4 rules. About 1 is common, 0 are distinctive, and 3 are very specific to this project.

  • ≈1

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    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:

    After writing any note in .idea/ or .management/, validate it, and fix what it reports:

    line 61
  2. Very specific:

    Never delete or overwrite a note in .idea/ — supersede it and link the replacement.

    line 68
  3. Common:

    Never fabricate a version, a CLI flag, a command's output, or a secret.

    line 69
  4. Very specific:

    Never edit .claude/ to make it fit your tool.

    line 70

≈ “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

AGENTS.md

For AI coding tools other than Claude Code — Codex, Cursor, Gemini CLI, Copilot, and anything else that reads AGENTS.md.

tateru is Claude-first. It is built for Claude Code, and the complete instructions live in CLAUDE.md. This file does not repeat them. It tells you what carries over to your tool, what does not, and how to follow the workflow by hand.

Read first

  1. CLAUDE.md — the project rules. They apply to you too, except where they name a Claude Code feature you do not have.
  2. .claude/skills/shared/vault-contract.md — how every note in .idea/ and .management/ must be written.
  3. .claude/skills/shared/quality-bar.md — how stack and structure get decided.
  4. .claude/team/stack.md, once /setup has written it — the stack, the real commands, and the specialist persona for each part of the codebase.

What works in any tool

PieceWhere
The plan: ideas, spec, roadmap, epics, decisions.idea/ — plain Markdown
Execution state.management/ — Markdown, state.json, board.db (SQLite)
Commit tracking and pre-push checks.githooks/, wired by git config core.hooksPath .githooks
Dashboard conformance checkbash .claude/scripts/dashboard-conformance.sh
The method behind every command.claude/skills/<name>/SKILL.md and its references/

What only works in Claude Code

  • Slash commands (/brainstorm, /shape, /setup, /work-on, …). Your tool will not run them.
  • Subagents in .claude/agents/, including the /work-on team and its parallel slices.
  • The vault hooks in .claude/settings.json. Nothing validates your notes as you write them, so check them yourself (below).
  • AskUserQuestion — where a skill uses it, ask the user in plain text instead.

Following the workflow by hand

Each command is a thin wrapper around a skill. To do what a command does, open its skill and follow it as a procedure:

CommandFollow
/brainstorm.claude/skills/brainstorm/SKILL.md
/shape.claude/skills/shape/SKILL.md
/epic.claude/skills/epic/SKILL.md
/setup.claude/skills/setup/SKILL.md
/work-on.claude/skills/work-on/SKILL.md — do every agent's step yourself, in order, without parallel slices
/ship.claude/skills/ship/SKILL.md
/maintain.claude/skills/maintain/SKILL.md
/upgrade.claude/skills/upgrade/SKILL.md
/decide.claude/commands/decide.md — it has no skill

Expect to be slower and less checked than in Claude Code. That is the cost of not being in the tool the workflow was built for.

Non-negotiables

  • After writing any note in .idea/ or .management/, validate it, and fix what it reports:

    bash .claude/hooks/validate-vault-note.sh --check "<path to the note>"
  • Application code lives in app/, and only /setup's procedure creates it.
  • Never delete or overwrite a note in .idea/ — supersede it and link the replacement.
  • Never fabricate a version, a CLI flag, a command's output, or a secret.
  • Never edit .claude/ to make it fit your tool. Claude Code users depend on it as it is.

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