Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/iroha924/mumei/composenpx skills add iroha924/mumei --skill composegit clone --depth 1 https://github.com/iroha924/mumeiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/iroha924/mumei/compose)<a href="https://agentmods.dev/skills/iroha924/mumei/compose"><img src="https://agentmods.dev/badge/skills/iroha924/mumei/compose.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00213 | $0.20436 |
| Opus 5 | $0.00106 | $0.10218 |
| Sonnet 5 | $0.00043 | $0.04087 |
| Haiku 4.5 | $0.00021 | $0.02044 |
Grade A, and why
compose scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 1,480 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Compose — mumei orchestrator
You orchestrate the full lifecycle of a feature in mumei: gather input → clarification → requirements → design → tasks → single user approval gate → implement (Wave by Wave) → 4-stage review → done.
This skill is the heart of mumei. Every other skill (glean, kindle, shelve) plays a supporting role.
Language conventions (applies to all spec drafts: requirements / design / tasks)
The skill produces three documents per feature: requirements.md, design.md, tasks.md. They follow a consistent language policy:
- Section headings stay in English, always. Hooks, parsers, the orchestrator, and the spec-reviewer agents rely on stable English headings (
## User Story,## Acceptance Criteria,## Out of Scope,## Architecture,## Wave 1: ..., etc.). Do not translate them. - Body content follows the user's conversation language. Detect the language from the user's recent substantive messages:
- If the user writes in Japanese, draft the User Story prose, the body of each AC (the clause after EARS keywords), Assumptions text, Open Questions text, design narratives, task descriptions, and Wave goals/verifies in Japanese.
- If the user writes in English, write everything in English.
- If mixed, default to the language of the user's most recent substantive message.
- EARS keywords stay in English regardless of body language:
WHEN,WHILE,IF,WHERE,SHALL. This keeps acceptance criteria machine-parseable. - Annotations stay in English:
[CONFIRMED],[ASSUMPTION],[NEEDS CLARIFICATION: ...]. These are read by therequirements-revieweragent. - Trace IDs stay as-is:
REQ-1.1,REQ-1.2, etc. - Task meta stays in English:
_Files:_,_Depends:_,_Requirements:_. The values inside are file paths and IDs (also unchanged).
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 1,480 lines · 213 tokens per session scan A 09cfcdd7e805
compose is a skill published in the GitHub repository iroha924/mumei (2 stars, last pushed 5d ago), licensed MIT. It adds 213 tokens to every session and 20,436 once invoked, about $0.0011 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
harness
하네스를 구성합니다. 전문 에이전트를 정의하며, 해당 에이전트가 사용할 스킬을 생성하는 메타 스킬. (1) '하네스 구성해줘', '하네스 구축해줘' 요청 시, (2) '하네스 설계', '하네스 엔지니어링' 요청 시, (3) 새로운 도메인/프로젝트에 대한 하네스 기반 자동화 체계를 구축할 때, (4) 하네스 구성을 재구성하거나 확장할 때, (5) '하네스 점검', '하네스 감사', '하네스 현황', '에이전트/스킬 동기화' 등 기존 하네스 운영/유지보수 요청 시 사용.
evaluator-write-qa-parallel
Internal Auto-Harness evaluator skill for parallel sprint QA and QA report writing. Use only inside the Evaluator subagent during evaluatorqaparallel.
evaluator-write-qa
Internal Auto-Harness evaluator skill for sprint QA and QA report writing. Use only inside the Evaluator subagent during qa mode.
evaluator-write-retest-parallel
Internal Auto-Harness evaluator skill for parallel sprint retest and retest report writing. Use only inside the Evaluator subagent during evaluatorretestparallel.
evaluator-write-retest
Internal Auto-Harness evaluator skill for sprint retest and retest report writing. Use only inside the Evaluator subagent during retest mode.
generator-fix-integrator
Internal Auto-Harness generator skill for worktree-merge fix integration. Use only inside the parallel Generator integrator when it is addressing named defects from QA or retest.