agentleFS
JoaoEquerJoaoEquer/Oficina/AGENTS.md

Oficina

AGENTS.md2 starsChanged 24 days agoMITView on GitHub

Entry point for harnesses that read AGENTS.md (Codex, Cursor, OpenCode, Gemini CLI and similar). This repository is a lean harness of working patterns for the Node.js + TypeScript + Prisma stack — Express (manual Clean Architecture) in the production backends it was extracted from, NestJS kept for projects that explicitly pick it. When working on any project that references this harness: Curated skills (audited external adoption closing a confirmed gap — Path 2 in docs/HOW-TO-GROW.md): - domain-modeling — building/sharpening CONTEXT.md…

  • Installs packages

The file

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

1 rule. About 0 are common, 1 is distinctive, and 0 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.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When working on any project that references this harness:

    line 7

≈ “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 — Oficina

Entry point for harnesses that read AGENTS.md (Codex, Cursor, OpenCode, Gemini CLI and similar).

This repository is a lean harness of working patterns for the Node.js + TypeScript + Prisma stack — Express (manual Clean Architecture) in the production backends it was extracted from, NestJS kept for projects that explicitly pick it.

When working on any project that references this harness:

  1. Follow the always-on rules in rules/:
    • rules/engineering-philosophy.md — how the team decides (pragmatism, no over-engineering)
    • rules/git-workflow.md — branches, commits, documentation from day 0
    • rules/security-baseline.md — security and multi-tenancy non-negotiables
    • rules/agent-workflow.md — how to direct the agent itself (plan before code, scoped sessions, early correction)
  2. Consult the skills in skills/ before executing tasks in the covered areas. House patterns (extracted from repeated real usage — Path 1 in docs/HOW-TO-GROW.md):

    • express-prisma-pattern — creating any CRUD route/domain in an Express + Prisma backend (the house stack)
    • prisma-schema-conventions — creating/changing schema, planning migrations
    • rbac-design — permissions, roles, access control
    • clickup-task-breakdown — breaking scope into tasks
    • client-facing-docs — any material the client will read
    • token-efficiency — session setup, token/context limits, RTK
    • design-audit — read-only audit for the generated-UI look and design-system drift (static scan + rendered check, ranked report)
    • frontend-design — pick a deliberate visual direction (plan, review against brief, build) before writing UI; pair with design-audit to check the result
    • break-ui — feeds a component worst-case data (long names, empty states, huge counts, non-Latin text) behind a demo toggle and reports what breaks, with a fix for each
    • security-audit — read-only code security audit, source-to-sink tracing, HIGH-confidence findings only (Express/Prisma, web, Electron, Expo, LLM)
    • ai-news — daily AI news briefing in PT-BR (AlphaSignal + Hermes AI, new items only) and Shorts scripts from the best stories; output in ~/.oficina/news/, never in a repo

    Curated skills (audited external adoption closing a confirmed gap — Path 2 in docs/HOW-TO-GROW.md):

    • domain-modeling — building/sharpening CONTEXT.md glossary and ADRs
    • codebase-design — deep-module vocabulary when designing or reviewing an interface
    • improve-codebase-architecture — scanning a codebase for deepening opportunities
    • grilling — structured requirements interview to stress-test a plan or decision
    • diagnosing-bugs — hard bugs and performance regressions
    • tdd — test-first feature/bugfix work, red-green-refactor discipline
    • secrets-gitleaks — secret scanning, pre-commit and CI (operationalizes security-baseline.md)
    • ci-cd-and-automation — standing up CI/CD quality gates from scratch
    • redesign-existing-projects — auditing and upgrading the UI of an existing project without changing its stack
    • screenshot-to-code — rebuilding a UI from a screenshot, then comparing the render against it
    • web-design-guidelines — reviewing front-end code against Vercel Web Interface Guidelines (rules vendored, no network call)
    • playwright-cli — driving a real browser from the terminal to test a page at phone width (needs npm i -g @playwright/cli)
    • rag-architect — designing/evaluating a RAG pipeline
    • sentry-setup-ai-monitoring — instrumenting LLM call cost/latency/tokens

    Motion (experimental até o primeiro vídeo real que usar):

    • motion-dictionary — vocabulário de 25 efeitos (entradas, ritmo, texto, formas, transições e princípios de animação) com duração, easing e parâmetros padrão, gramática de spec ("peça assim"), 5 modelos de prompt em blocos e fx.js determinístico (função do tempo, sem dependências) para HTML-to-video, GSAP, Remotion, CSS ou Framer Motion

    Game projects only (experimental — not yet used in two real projects; load only when the project is a game, never in backends or web apps):

    • game-design-doc — building/reviewing a GDD (pillars, core loop, economy, scope); /oficina:gdd
    • game-assets — asset layout, naming and the AssetForge viewer/export scaffold for Unity, Electron and Roblox; /oficina:asset-forge, /oficina:asset
  3. Documentary source of truth: data model and scope come from the project's documents. Divergence is a blocker to resolve with the tech lead, not a detail to ignore.

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.