agentleFS
Sign inStart free
Alex5350Alex5350/arsenal/AGENTS.md

arsenal

AGENTS.md0 starsChanged 16 days agoMITView on GitHub

You are a coding agent working in arsenal, a provider-agnostic AI development hub. This file is the universal entry point: Claude Code, Codex CLI, OpenCode, and GitHub Copilot all read it. The full knowledge base lives in docs/. arsenal is a template repository teams clone so that every AI coding harness shares one source of truth: skills, engineering standards, model routing policy, and work tracking. It is stack-agnostic and carries no application code of its own. Consumer repositories keep their…

  • 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 31,439 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 repository7 other files it gives its agents.

The file

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

4 rules. About 0 are common, 2 are distinctive, and 2 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.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    Track all work: Every unit of work is mirrored to the team tracker (GitHub Issues, Jira, or Azure DevOps) using docs/integrations/ and …

    line 24
  2. Very specific:

    Never commit secrets or credentials.

    line 53
  3. Distinctive:

    Never disable or skip a test to make progress.

    line 54
  4. Distinctive:

    When a harness default or a stale habit conflicts with this repository, this repository wins.

    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

AGENTS.md

You are a coding agent working in arsenal, a provider-agnostic AI development hub. This file is the universal entry point: Claude Code, Codex CLI, OpenCode, and GitHub Copilot all read it. The full knowledge base lives in docs/.

What this repository is

arsenal is a template repository teams clone so that every AI coding harness shares one source of truth: skills, engineering standards, model routing policy, and work tracking. It is stack-agnostic and carries no application code of its own. Consumer repositories keep their own tech stack; arsenal governs how agents work in it.

Operating principles

  1. One hub, many harnesses. docs/ is canonical. Never duplicate guidance into a harness-specific file; root adapter files only point into docs/.
  2. Follow the loop. Research, plan, execute, verify, track, publish. Phase definitions and gates live in docs/workflows/.
  3. Route by tier. Strategist models plan and review; executor models write code. The current matrix and per-harness wiring live in docs/models/routing.md.
  4. No handoff without green. Work moves to the next phase only after the test gate defined in docs/standards/testing.md passes.
  5. Track all work. Every unit of work is mirrored to the team tracker (GitHub Issues, Jira, or Azure DevOps) using docs/integrations/ and scripts/work-item.sh.
  6. Skills are the unit of reuse. Procedural knowledge goes in .agents/skills/ as Agent Skills (SKILL.md, the open standard at agentskills.io); role agents go in .claude/agents/. Every harness reads those folders where they are (Claude Code through the committed plugin in .claude-plugin/), so a new skill needs no sync step. A new agent also needs its opencode.json entry and make sync for .codex/agents/<name>.toml, the only generated files; never edit those by hand, CI fails when they drift. Authoring rules: docs/skills/README.md and docs/agents/README.md.

Where things live

You needRead
The full hub map and reading orderdocs/README.md
How your harness consumes this repodocs/harness/
Engineering standardsdocs/standards/
The six-phase delivery loopdocs/workflows/
Model tiers and routingdocs/models/routing.md
Shared skills library.agents/skills/ (rules: docs/skills/)
Role agents (researcher, planner, reviewer).claude/agents/ (contract: docs/agents/)
Work tracking backendsdocs/integrations/
Superpowers methodology layerdocs/superpowers/

Hard rules

  • Never commit secrets or credentials. See docs/standards/security.md.
  • Never disable or skip a test to make progress. Fix the code or fix the test.
  • Conventional commits only. See docs/standards/git.md.
  • Stamp work-item IDs in branch names, commits, and pull requests.
  • When a harness default or a stale habit conflicts with this repository, this repository wins.

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