group-discussion-reviewer-methodology
Skill1 starsChanged 3 months agoMITView on GitHub
用一套严格的"产品组会评审"方法论,在对话里直接评审一份 PRD —— 无需 API key、无需运行任何代码。产出双轴结论(组内准入 / 模拟评审结果)+ P0/P1/P2 问题,并对 P0 严格把关(只有真正阻断审批/安全上线/可衡量/方向/交付的问题才算 P0)。当用户要"评审 PRD / 模拟组会评审 / 上会模拟 / 找 P0 阻断项 / 组内准入判断"但没有可用的 LLM key、或更想要一个可照做的思维框架时使用。English triggers: "review this PRD", "simulate a group/product review", "find P0 blockers", "is this PRD ready", "use the group-discussion review framework".
In this repository12 other files it gives its agents.
- agent-fleet-guardSkill
- audience-briefSkill
- blind-ab-verifySkill
- change-triageSkill
- codex-review-gateSkill
- group-discussion-reviewerSkill
- lark-comment-loopSkill
- lark-peer-feedback-draftingSkill
- meeting-notesSkill
- metric-change-attributionSkill
- session-handoffSkill
- skill-evolve-loopSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
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
- Very specific: line 23
Read methodology.md in this folder and execute its 8 phases in order.
- Very specific: line 41
Always read methodology.md before reviewing.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill 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.
Files
Group Discussion Reviewer — methodology (no key, inline)
Apply a disciplined product group-discussion review to a PRD directly in the conversation. No API key, no Node, no code execution — you (the agent) perform the review by following the written methodology.
This is the lightweight sibling of the
group-discussion-reviewerengine skill. The engine runs the same pipeline deterministically (and gates P0s with code) but needs an OpenAI-compatible key. Prefer the engine when a key is available and fidelity matters; use this methodology skill when there's no key, or for a quick framework-driven review.Honest limitation: without the engine's deterministic P0 gatekeeper, the strictness of P0 downgrading depends on your judgment. Hold the P0 bar hard.
How to use
- Get the PRD body as text/markdown (this skill does not fetch URLs or expand sheets).
- Read
methodology.mdin this folder and execute its 8 phases in order. - Produce exactly the output contract:
## 组内准入,## 模拟评审结果,## P0 Blockers,## P1 Questions,## P2 Questions,## P1/P2 Improvement Suggestions. - Respect the mode the user asks for (standard / deep / challenge; default challenge) and the language (default Chinese).
The core discipline (summary — full detail in methodology.md)
- Two independent axes: 组内准入 (is it concrete enough to discuss?) vs 模拟评审结果 (would it pass?). A PRD can be admitted and still 不通过.
- P0 only for four failures: wrong direction/population, unmeasurable success (for a formal decision), unsafe launch, or non-viable delivery — each with a concrete product bridge. Everything else (stats audit, SLA/QPS/monitoring, tracking schema, release checklist, compliance checklist, naming/icons, missing sections) defaults to P1/P2.
- A clean PRD has 0 P0. Don't inflate P0 to fill a quota.
- Blind review: PRD body only; ignore comments/annotations/post-meeting context. Ask questions, don't answer them — except every P0 must include a concrete revision.
Always read methodology.md before reviewing; it carries the full P0 gate rules, the
calibration priors, and the reviewer lenses.
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
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.

