agentleFS
wrsilvawrsilva/reis-mobile/AGENTS.md

reis-mobile

AGENTS.md8 starsChanged 58 days agoMITView on GitHub

Instructions for AI agents (Claude Code, Codex, Gemini) working in this repository. reis-mobile is a Claude Code plugin with a Node.js CLI that orchestrates agents and skills for mobile development. Read ARCHITECTURE.md before changing the core. Do not improvise a release. package.json is the only version you edit, and a bump without notes under ## [Unreleased] in CHANGELOG.md fails CI. The manifests, the dated changelog heading and the README release block are generated by npm run version:sync; the tag, the…

The file

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

2 rules. About 0 are common, 1 is distinctive, and 1 is 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.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Mobile first: Every new agent, skill or command must directly help mobile development.

    line 12
  2. Very specific:

    Every agent and skill declares intents and stacks in the frontmatter, or routing: manual.

    line 19

≈ “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

Instructions for AI agents (Claude Code, Codex, Gemini) working in this repository.

Project

reis-mobile is a Claude Code plugin with a Node.js CLI that orchestrates agents and skills for mobile development. Read ARCHITECTURE.md before changing the core.

Rules

  • English only. Agents, skills, commands, documentation, templates, comments and commit messages are written in English. Portuguese stays only where it is functional: the intent vocabulary in core/router/intents.mjs and the tests that cover Portuguese prompts.
  • Mobile first. Every new agent, skill or command must directly help mobile development.
  • Zero dependencies. core/ and bin/ use only the Node.js 22+ standard library. Do not add npm packages.
  • ES modules (.mjs), 2 spaces, single quotes, semicolons.
  • Do not invent. Skill checklists cite APIs, lints, flags and store requirements that actually exist. When in doubt, tell the model to check the version in the project's lock file instead of asserting it.
  • Every skill starts with mobile-, and there are two kinds. A topic skill (mobile-test, mobile-debug, mobile-firebase...) covers every stack with stacks: ["*"]: SKILL.md holds what is true everywhere and a table pointing to references/flutter.md, android.md, ios.md, react-native.md and kotlin-multiplatform.md. A platform skill (mobile-flutter, mobile-android, mobile-ios, mobile-rn, mobile-kmp) holds what exists only on that stack, grouped by theme in references/<theme>.md. A reference with a lot of material becomes an index plus a folder (references/android.md → references/android/<subtopic>.md), so the model loads only what the task needs. Add to an existing skill before creating a new one. reis-mobile validate enforces the names and stacks, and tests check that every platform reference is linked and every relative link resolves.
  • Focused references. A reference solves one problem (references/ios/cocoapods.md, references/flutter/riverpod.md); a skill is a topic or a platform, not a whole domain (mobile-development).
  • Third-party skills are added as references of the matching reis-mobile skill: their instructions stay as upstream wrote them, with an attribution line at the top (> Adapted from the \<name>` skill in owner/repo (License).), whole folders when they ship extra files (GUIDE.mdplus their references and scripts), and an entry inTHIRD_PARTY_NOTICES.md` with the license text. A test fails when an adapted file is not credited.
  • Every agent and skill declares intents and stacks in the frontmatter, or routing: manual.
  • Changes to the detector, the router or the redaction require a test in tests/.

Verification

npm run check    # validate + test

Releasing

Do not improvise a release. package.json is the only version you edit, and a bump without notes under ## [Unreleased] in CHANGELOG.md fails CI. The manifests, the dated changelog heading and the README release block are generated by npm run version:sync; the tag, the npm package and the GitHub release are created by the Release workflow, which needs two merges. The process, the repository settings it depends on and every failure mode are in docs/releasing.md.

Adding components

I want...CreateAnd also
Guidance for a topic that exists on several stacksreferences/<platform>.md in the matching skills/mobile-<topic>/, or a new mobile-<topic> with stacks: ["*"]Link it from SKILL.md
Guidance that exists on one stack onlyreferences/<theme>.md in skills/mobile-flutter, mobile-android, mobile-ios, mobile-rn or mobile-kmpLink it from the skill's topic table
An agentagents/<name>.mdA routing test for the intent
A commandcommands/<name>.md (becomes /reis-mobile:<name>)Documentation in the README
An intentTerms in core/router/intents.mjsCases in tests/intent-detector.test.mjs
A stackRule in stack-detector.mjs + stacks/<id>/stack.jsonCases in tests/stack-detector.test.mjs
An MCP toolAn entry in TOOLS in core/mcp/tools.mjs, read-only and offline, returning a payload from core/commands/payloads.mjsCases in tests/mcp.test.mjs and a row in the README tools table

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.