mpc_implement-spec
Make the results of a whole spec from /boarlabs:mpc_to-spec and /boarlabs:mpc_to-tickets: every ticket's result on its platform, each one checked by /boarlabs:boar_review.
- License 许可
- MIT
- 最近更新
- 2026-10-21
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- mpc_implement-spec
- description
- Make the results of a whole spec from /boarlabs:mpc_to-spec and /boarlabs:mpc_to-tickets: every ticket's result on its platform, each one checked by /boarlabs:boar_review.
- disable-model-invocation
- true
03 SKILL.md 正文
You have been provided a spec. This spec should have tickets associated with it, describing what to make.
The issue tracker should have been provided to you. If not, tell the user to run /boarlabs:boar_setup.
The goal is every ticket's result made on the platform its ticket names (a Google Doc or Drive file, a ClickUp task or doc, a Miro board, a Claude artifact), checked, and every ticket resolved the way the issue tracker closes work. Results never go into git: there is no integration branch, no worktree and no pull request.
The tickets are not a list of steps. They are a task graph with blocking relationships between them. This means there is always a frontier of tickets which are ready to be grabbed.
Communication to and from subagents should be sparse. Communicate primarily through context pointers: to the spec, tickets, research notes and results already made. Don't duplicate information already available via pointers.
Maker subagents should be run in the background where possible for maximum concurrency.
Steps
-
Read the spec and tickets to understand the task graph. Every ticket must carry a Result section (platform, place with a link, name) and a Marketing line (
yesorno). If one lacks either, stop and ask the user; there is no default. -
(optional) When the tickets need exploration (a question about the web or about the repository's context), call the Skill tool with
boar_codexand start a research run with a precise question and a criterion of done. Read its evidence before relying on it. Save what it found as notes the maker subagents can point at. -
Use maker subagents to make each ticket on the frontier, one subagent per ticket. Each maker subagent:
- reads its ticket, the spec and the notes it is pointed at;
- makes the result on the ticket's platform, in the ticket's place, under the ticket's name, through the session's connectors;
- with
Marketing: yes, uses thechs_*skills that fit the result, picked by their descriptions; withMarketing: no, uses none of them; - reports the result's link, and nothing else changed.
-
Check each finished result one at a time, in this session: call the Skill tool with
boar_reviewfor the ticket. Playwright drives one browser, so two checks never run at once. Meet its findings on the platform, asboar_reviewsays. -
Close each checked ticket the way the issue tracker closes work, with a comment that links the result and carries the
boar_reviewreport. -
If this changes the frontier of available tickets, kick off more maker subagents for the new tickets. This allows for maximum concurrency.
-
Commit and push what this run wrote to the repository (notes, glossary or ADR changes), as described below. Report every ticket with its result's link.
Committing
Results live on their platforms and are never committed. What this run wrote to the repository itself is context: commit only those paths (git add <paths>, never git add .) on main with a short message naming the spec, then git push. If the push is refused, stop and tell the user why; never force it.
04 同仓库更多 Skills
mpc_writing-shape
Writing, exploit: shape raw material into an article, paragraph by paragraph.
mpc_writing-fragments
Writing, explore: mine raw fragments, no structure yet.
mpc_writing-for-agents
Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.
mpc_writing-beats
Writing, exploit; assemble raw material into a journey of beats, grounding each term before a beat leans on it.
mpc_wayfinder
Plan a huge chunk of work (more than one agent session can hold) as a shared map of decision tickets on your issue tracker, and resolve them one at a time until the way to the destination is clear.
mpc_wait-what
Stop. That last message did not land: re-pitch it.