agentleFS
dawn840705dawn840705/codex-code-studios/CLAUDE.md

codex-code-studios

CLAUDE.md17 starsChanged 2 months agoMITView on GitHub

You are in the source repository of a Codex plugin. This tree builds codex-code-studios; the studio it ships runs inside Codex, invoked as $skill-name. AGENTS.md is the source of truth for this repository — read it before changing anything, and prefer it over this file wherever the two could be read differently. This file exists because Claude Code loads it automatically. Its job is to stop you acting on assumptions carried in from the sibling package, and to hand you…

We split this CLAUDE.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
  1. Very specific:

    Read production/runtime.txt in the user's project — one line: codex (full studio, the default here) · claude (stand down, the sibling owns …

    line 33
  2. Very specific:

    When a script can judge something, the script judges it.

    line 57
  3. Distinctive:

    Before release, run the full check set AGENTS.md § Change rules 5 lists — the plugin validator, every skill validator, the skill linter, …

    line 66
  4. Distinctive:

    Do not re-derive these; they are written down.

    line 72

≈ “Common” is an estimate: at least 1 in 5 of the 25,680 CLAUDE.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
CLAUDE.md

codex-code-studios — guidance for Claude Code

You are in the source repository of a Codex plugin. This tree builds codex-code-studios; the studio it ships runs inside Codex, invoked as $skill-name. AGENTS.md is the source of truth for this repository — read it before changing anything, and prefer it over this file wherever the two could be read differently.

This file exists because Claude Code loads it automatically. Its job is to stop you acting on assumptions carried in from the sibling package, and to hand you the right entry points.

What is NOT here

The Claude-side studio does not exist in this tree. Specifically:

You might assumeActually
45 agents in agents/No agents/ directory. The 45 role guides are skills/studio-orchestrator/references/roles/*.md — references the orchestrator loads, not installed agents. Do not spawn them as subagents.
Skills invoked as /nameCodex invokes $name. Write $skill-name when referring to an invocation.
.claude/settings.json holds project configProject-owned config is .codex/studio/technical-preferences.md; layout overrides are .codex/studio-layout.json.
.claude-plugin/ and this file are currentBoth are a transition-compatibility layer kept for the 0.7 release (AGENTS.md). .claude-plugin/plugin.json is a frozen v0.6.4 copy still named claude-code-studios. Neither is a source of truth, and neither should be ported back to the sibling.

The sibling package claude-code-studios (Claude Code, /skill-name, a real agents/ tree) is a separate repository. Nothing here installs it.

Which runtime is working the project

This studio ships twice, and a user project may be worked by Codex alone, by Claude Code alone, or by both at once. That choice changes whether this plugin acts at all.

Read production/runtime.txt in the user's project — one line: codex (full studio, the default here) · claude (stand down, the sibling owns the work) · split (work only your partition, never a path the other runtime owns). Absent means codex. Never infer it from a CLAUDE.md or .claude/ sighting — that proves Claude Code was configured there once, not that it is working this task.

Full rule, including why split risks a silently lost edit: rules/runtime-modes.md.

작업 원칙

이 절은 프로젝트 정보가 아니라 기본 동작 교정이다. 지시하지 않으면 반대로 가는 것만 적는다.

  • 해결책을 설계하기 전에, 이미 자리 잡은 제품들이 같은 문제를 어떻게 푸는지 먼저 살펴보세요. 접근 방식을 처음부터 발명하지 말고 검증된 패턴과 관례를 채택하세요.
  • 스킬 이름·산출물 경로 규약의 하위 호환을 깨지 마세요. 사용자가 설치하는 배포물이고 이것들이 사실상 공개 API입니다. 바꿔야 한다면 새 경로를 추가하고 기존 경로는 CHANGELOG에 deprecated로 표시한 뒤 최소 한 마이너 버전 유지하세요. 반면 플러그인 내부 구현(스크립트 헬퍼, 훅 lib, 템플릿 내부 구조)에는 호환 레이어를 쌓지 말고 쓰이지 않는 경로를 삭제하세요.
  • 현재 요구사항을 완전히 충족하는 가장 단순한 구현을 선택하세요. 추측에 근거한 추상화, 설정값, 간접 계층을 만들지 마세요.
  • 시스템은 레이어로 키우세요. 엔드투엔드로 동작하는 최소 버전에서 시작하고, 이미 동작하는 결과물 위에 기능을 하나씩 얹으세요. 동작하는 코드를 미완성 복잡도와 맞바꾸지 마세요.
  • 컴포넌트는 모듈로 분리하고 관심사를 명확히 나누세요.
  • 검증된 라이브러리가 전체 복잡도를 낮춘다면 그것을 쓰세요. 단 이 저장소의 런타임은 표준 라이브러리만 씁니다 — 의존성 추가는 별도 승인 사항입니다 (docs/deterministic-gates.md "Adding a gate" 규칙 5).
  • 직접 구현하거나 패키지를 추가하기 전에 이미 설치된 의존성부터 확인하세요. 문서와 타입을 확인하지 않은 채 "이 라이브러리엔 그 기능이 없다"고 단정하지 마세요.
  • 아키텍처 결정은 장기 관점으로 하세요. 지금만 넘기고 나중에 교체할 임시방편을 받아들이지 마세요.

The exit code is the verdict

When a script can judge something, the script judges it. Exit codes are the contract: 0 pass · 1 warning · 2 abort · 3 could not run. A gate that could not run has produced no verdict — never read that as a pass. Do not re-derive a verdict by parsing a runner's output; read the exit code, then use the text only to explain it. Name the deciding gate in every verdict: pytest → exit 1 (3 failed).

Contract and how to add a gate: docs/deterministic-gates.md.

Before release, run the full check set AGENTS.md § Change rules 5 lists — the plugin validator, every skill validator, the skill linter, shell syntax checks, and the test suite.

Other rules that apply to you here

Do not re-derive these; they are written down.

QuestionRead
Iterating on a quality-gated deliverablerules/self-loop.md
How many workers to spend on a taskrules/route-hint.md
How hard to verify a changerules/verify-route.md
Full rule indexdocs/rules-reference.md
Repository layoutdocs/directory-structure.md

Do not pre-create empty directories. Skills create what they need; an empty tree makes phase detection read a fresh project as one that already started.

docs/ holds everything else — list it rather than trusting an index carried in a guide file.

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.