메뉴
Explore codebase with Haiku, synthesize a tight machine-readable implementation spec. Output feeds directly into /feature-coder or /tdd-fixer — no plan-audit loop. Triggers on: /spec-writer, write a spec, spec out this feature, spec before coding.
기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.
설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
직접 명령은 검토 Prompt를 거치지 않습니다. 실행하기 전에 소스를 확인하세요.
npx skills add https://github.com/dkorduban/twilight-struggle-ai --skill spec-writer명령은 한 줄로 유지됩니다. 복사하기 전에 가로로 스크롤해 전체 내용을 확인하세요.
로컬 사본을 원하시나요? SkillsMP에서 현재 제공할 수 있는 파일을 다운로드하세요.
How to use OpenAI Codex via MCP tools for code generation, plan review, and collaborative implementation loops. Use when calling Codex for implementation, review, or any cross-model collaboration.
Master task dispatcher. Simple tasks (≤2 files, ≤30 lines, clear spec) are done inline by the main agent. Complex tasks are routed to bg-codex-implementer as a background agent. Triggers on: /dispatch, implement this, fix this, debug this, dispatch task.
Benchmark one or more model checkpoints (2000 games/side) with auto-logging. Triggers on: /bench, benchmark this model, run benchmark, bench checkpoint.
SOC 직업 분류 기준
SKILL.md 표시 중
| name | spec-writer |
| description | Explore codebase with Haiku, synthesize a tight machine-readable implementation spec. Output feeds directly into /feature-coder or /tdd-fixer — no plan-audit loop. Triggers on: /spec-writer, write a spec, spec out this feature, spec before coding. |
Produces a .claude/plan/<name>.md spec ready for /feature-coder or /tdd-fixer.
No Codex audit loop. No Plan agent. Haiku does discovery; this session synthesizes.
Task description as the argument. Optionally: an existing rough notes file.
Phase 1 (Haiku Explore subagent) does all file reading. This session only synthesizes. Target: ≤ 5 main-session turns on the happy path.
Task — Haiku Explore subagent for discoveryRead, Glob, Grep — fallback lookups onlyWrite — save spec fileLaunch a Task agent (subagent_type: "Explore", model: "haiku") with:
Explore the codebase to answer these questions for the task: "{task description}"
1. Which existing files are most relevant? (list paths + 1-line purpose)
2. What test file conventions exist? (naming, fixture patterns, imports)
3. What interfaces/types/schemas does the new code need to conform to?
4. Are there similar existing implementations to follow as a pattern?
5. What are the exact file paths that need to be created or modified?
Return structured answers only. No prose. No implementation.
Using the Explore output, write .claude/plan/<feature-name>.md with this structure:
# Spec: <feature name>
## Goal
One paragraph. What this does and why.
## Files to create
- `path/to/new_file.py` — purpose
## Files to modify
- `path/to/existing.py` — what changes and why
## Interfaces / signatures
Exact function signatures, class definitions, schema fields.
Copy from existing patterns where applicable.
## Test cases (required)
List as bullet points:
- test_<name>: <what it asserts> — <setup required>
- test_<name>_edge_case: <edge condition>
(minimum 3, cover happy path + 1 error path + 1 edge case)
## Acceptance criteria
- [ ] All listed test cases pass
- [ ] No regressions: `uv run pytest tests/python/ -q` green
- [ ] Follows existing code conventions (no new deps, no style changes)
## Constraints
Any project-specific rules (e.g. "no in-place frozenset mutation", "use _vp_delta not pub.vp +=")
Print:
Spec saved to `.claude/plan/<feature-name>.md`
To implement: /feature-coder .claude/plan/<feature-name>.md
To TDD: /tdd-fixer .claude/plan/<feature-name>.md
Stop. Do not implement.