Skip to content

majiayu000/spellbook / plugins/idea-coach/skills/idea-team

idea-team

Verified

想法群聊室主持人 — 把一句话想法丢给多角色 AI 团队(调研员/反方/类比者)做查漏补缺。每个角色有自己的 voice,他们互相 @ 接话;你随时插话。**这是创意扩展工具,不打分、不否决、不堵路**。Use when 用户说"组个团队聊一下"、"开会讨论这个想法"、"找几个角度看看"、"群聊一下 X"、"team review X",或调用插件命令 `/idea-coach:idea-team`。Do NOT use when 用户已决定做这个想法只要 PRD(用 `idea-to-product`)、或用户要单独某个角色 skill、或用户只想自由 brainstorm(不需要 team 结构)。

FileKindSize
SKILL.mdskill md9.5 KB
evals/evals.jsonconfig2.6 KB

evals/evals.json

View on GitHub
{
  "skill_name": "idea-team",
  "evals": [
    {
      "id": 1,
      "prompt": "用 idea-team 讨论:做一个本地 AI 账单分析器。",
      "expected_output": "Emits all three role blocks in one assistant response, persists canonical utterances and speakers atomically, projects each utterance once to chat, then invites user input.",
      "files": [],
      "expectations": [
        "Researcher, devil's advocate, and analogist appear as separate ordered blocks in one response.",
        "Each utterance has a stable round:role id in state.",
        "round_complete is written with the third speaker rather than in a later independent write."
      ]
    },
    {
      "id": 2,
      "prompt": "恢复 round=2、speakers=[research]、round_complete=false 的群聊。",
      "expected_output": "Reconciles state and chat markers, emits only the missing two roles, persists each canonical utterance, and derives round_complete without repeating research.",
      "files": [],
      "expectations": [
        "The existing research utterance is not repeated.",
        "Missing chat projections are repaired idempotently.",
        "round_complete is derived from the complete speaker set."
      ]
    },
    {
      "id": 3,
      "prompt": "上一轮已完成,我说下一轮。",
      "expected_output": "Increments and persists the round with empty speakers before speaking, then completes all three role blocks and atomically marks the new round complete.",
      "files": [],
      "expectations": [
        "The round transition is persisted before the first new utterance.",
        "All three new utterance ids use the new round number.",
        "The user is invited to respond after exactly one round."
      ]
    },
    {
      "id": 4,
      "prompt": "同一个想法再开一次群聊,但 .idea-team/<slug>/ 已有 state.json 和 chat.md。",
      "expected_output": "Does not overwrite or append; asks to resume or uses a confirmed unique suffixed slug before creating state.",
      "files": [],
      "expectations": [
        "Existing state and chat remain untouched before the user chooses.",
        "A new session uses a unique slug such as <slug>-2."
      ]
    },
    {
      "id": 5,
      "prompt": "群聊已经写完 team-report.md。现在恢复这个会话。",
      "expected_output": "Persists status completed and completed_at when the report is generated, then resume only displays the report path.",
      "files": [],
      "expectations": [
        "Report generation atomically transitions state.status to completed.",
        "Resume does not start or repeat a round for a completed session."
      ]
    }
  ]
}
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall