prd
Skill0 starsChanged 34 days agoMITView on GitHub
把一句诉求变成可验证的契约五字段(goal / acceptance / invariants / unknowns / human_checkpoint):只问真正阻塞的澄清,其余自己定并标 [Assumption];落盘即视为可用,默认继续 /to-issues。**用户只给了一句诉求、落地方案还要自己定(「怎么落地你定」「做个能缓存的 KV」)时也走这里**——别直接开写。Triggers: 写PRD, 需求文档, 需求分析, plan this feature, 新功能, 需求还没成形, 落地方式你定.
125 other projects use a version of this.
Adopted by
Most recently taken up by
- Adelie-Squad/solosquadtheir own version, first seen Nov 8, 2026
- shaqmughal/seekstonetheir own version, first seen Nov 8, 2026
- lhbsaa/apex-unifiedtheir own version, first seen Nov 8, 2026
- Jasenpan1987/multi-codetheir own version, first seen Nov 8, 2026
- joeldevz/skynex-pitheir own version, first seen Nov 8, 2026
- insight68/Skillstheir own version, first seen Nov 8, 2026
Used with
Projects with it also use context-optimize, usecase and 6 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
ibahgatSkillcontext-optimize
Improve context efficiency and cache stability for long or complex sessions.
Also in 3 projects that use this
gabriellstSkillusecase
Create an application use case. Use when implementing business operations like CreateOrder, ProcessPayment, CancelBooking. Use this skill for any command/mutation that orchestrates domain logic, whether simple CRUD or complex multi-step transactions with sagas. Dispatch hub — routes to typescript/go variants by file extension.
Also in 3 projects that use this
ulises-jeremiasSkilltrd
WHAT — Draft and review a Technical Requirements Document (TRD) using the template, typically
Also in 3 projects that use this
snarktankSkillralph
Convert PRDs to prd.json format for the Ralph autonomous agent system. Use when you have an existing PRD and need to convert it to Ralph's JSON format. Triggers on: convert this prd, turn this into ralph format, create prd.json from this, ralph json.
Also in 16 projects that use this
JokerRunSkillcompound-engineering
Compound Engineering workflow for AI-assisted development. Use when planning features, executing work, reviewing code, or codifying learnings. Follows the Plan → Work → Review → Compound loop where each unit of engineering makes subsequent work easier. Triggers on: plan this feature, implement this, review this code, compound learnings, create implementation plan, systematic development.
Also in 3 projects that use this
chusiminSkillbrd
评估具体产品或业务方向是否值得投入,交付商业判断或 BRD。基于用户问题、替代方案、商业机制、成本和证据提出建议;按请求决定简短分析或完整文档。
Also in 3 projects that use this
Other versions125 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 125
- github/awesome-copilot★ 41ksame name
- snarktank/ralph★ 22ksame name
- SethGammon/Citadel★ 927same name
- smallnest/autoresearch★ 571same name
- snarktank/compound-product★ 542same name
- CatCatUncle/openworkbuddy★ 398same name
- smallnest/goal-workflow★ 283same name
- OffchainLabs/arbitrum-docs★ 252same name
- julianromli/ai-skills★ 215same name
- golbin/agent-skills★ 214same name
115 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository28 other files it gives its agents.
- AGENTS.mdAGENTS.md
- evals/AGENTS.mdAGENTS.md
- conflictSkill
- diagnoseSkill
- harness-engineeringSkill
- modern-goSkill
- open-source-skillSkill
- refactorSkill
- test-firstSkill
- triageSkill
- understandSkill
- graphSkill
- loop-itSkill
- merge-itSkill
- review-itSkill
- ship-itSkill
- to-designSkill
- to-issuesSkill
- find-skillsSkill
- frontend-designSkill
- humanizer-zhSkill
- pptxSkill
- resume-optimizerSkill
- skill-creatorSkill
- svg-diagramSkill
- teachSkill
- ui-ux-pro-maxSkill
- web-design-guidelinesSkill
The file
Files
prd — 一句诉求 → 可验证的契约五字段
/prd 把一句诉求变成工作状态的契约五个字段:goal / acceptance / invariants / unknowns / human_checkpoint。
工作状态、证据层、三个 profile 的边界、产物落点见 ../loop-it/CONTRACT.md——本文件不复述。 这里只写三件事:何时调用、产物长什么样、失败怎么办。
何时调用
- 用户给了一句诉求,落地方案还要自己定(「怎么落地你定」「文件格式你定」)——就是这里,别直接开写。
- 需求还没成形:写不出可证伪的验收条件,或作用域边界还没划。
- 「装得进一个上下文」不等于「定义好了」(CONTRACT §3)。小不是跳过规划的理由;反之,已经有人写好了验收条件(一条 issue 卡、spec 里的一项)就不需要 PRD,直接进
/loop-it。
输入
一句诉求 + 仓库上下文:AGENTS.md 的路由表、相邻代码、既有约定。能自己读到的就自己读,别拿它当澄清问题问回去。
输出
一份短文档,落在 <scope>/documents/prd-<feature>.md(作用域根见 CONTRACT §8;目录不存在就 mkdir -p):
- 没有远端 /
gh没登录时,这份文档就是契约的载体。 - 有
gh且人指定了 issue 时,直接写进那条 issue 正文——同一套字段,换个载体。
只写契约五字段。 谁受益、为什么现在做由 goal 承载,不必另起一节背景。简介 / 用户故事 / 功能需求 / 成功指标 / 技术考量都是按需的:有真实内容就写,没有就不写。长度由未知量决定,不由模板决定——九节结构不进默认路径。
| 字段 | 写什么 |
goal | 这次要改变什么、谁受益、为什么现在做(一句话,可证伪) |
acceptance | 验收条件;每条都要点出证明它为假的观测 |
invariants | 不能破坏什么 |
unknowns | 还不知道什么;自己定的决策标 [Assumption] |
human_checkpoint | 必须由人拍板的分叉(默认:无) |
acceptance 的判断:
- 「工作正常」「体验良好」不是验收条件。可观测的 UI 状态 / API 响应、清晰的输入输出对、能跑的工具检查才算。
- 按「能不能被同一次观测证明」合并,不按功能点逐条列。 想少写测试,杠杆是减少验收条件的条数(CONTRACT §4)。
- 有界面时,外观是一等验收条件——写进
acceptance(附视觉验证),不要划进非目标:把「好看」划出去,等于允许交付一个能用但没法看的东西。 - 每条要哪一层证据(L1–L4)执行时才标,不写进文档(CONTRACT §4)。
unknowns 是一等公民,不许静默猜: 每条只有两种归宿——有人认领,或显式写下「我按 X 假设推进」([Assumption])。Open questions: None 是最危险的一行,它通常是没想,不是没有。
一份填好的短 PRD 与带字母选项的提问格式见 references/example.md。
澄清
只问真正阻塞的:不回答就写不出可验证的 acceptance,或者答案会改变作用域边界。其余一律合理默认值 + [Assumption]。
- 用
ask_user_question,不要用散文把回合停在「等你回复」上——DSH 里turn_end: completed就是这个回合的终点。 - 问题带字母选项,人可以回「1A 2C」;数量随歧义伸缩,不凑条数。
- 能问到人时就在能问到人的上下文里做:子代理发起的
ask_user_question会被拒,所以别把这个技能派给子代理。 - 为什么不设评审闸门: PRD 是活文档,不是批准件;审批闸门把澄清切成瀑布。要评审就由
/review-it派不共享上下文的子代理做,人只在human_checkpoint被叫一次(CONTRACT §5)。
边界
- 不做实现。 只产出契约;落盘即视为可用,不等「回复 OK」。
- 不做第二份真相。 文档与工作状态冲突时以工作状态为准(CONTRACT §1);能只从状态 + issue 正文 + 代码推出来的东西,不用写。
- 不把「建议下一步」当产出:落盘后跑一次
route(见下),照它打印的next:加载下一个技能。 - 只在两种情况停:人明确说了「等评审过了再往下」,或「我只要文档」。
失败怎么办
| 场景 | 处理 |
工具列表里没有 ask_user_question | 无人值守部署。自己定:每个决定写进文档并标 [Assumption],照常产出 |
调用被拒:human interaction is unavailable while the calling agent is owned by another live agent | 你是子代理。同上——别停,把决定与未决项一起写进结果 |
| 人已经说了「怎么落地你定」 | 决定权已经交出来了。再拿落地细节回去问,是把决定权还回去——那不是澄清,是卡住 |
| 输入太含糊(「加个功能」) | 问一次;还是含糊就从项目上下文推断,标 [Assumption],落盘时点名这些假设 |
| 用户后来要改 | 当修订:改文档、回流到受影响的 issue 卡,不重跑澄清 |
| 拆到一半发现整个改动装得进一个上下文 | 直接说明,就地做完——不需要 issue,也不需要批次仪式(CONTRACT §3) |
停在「等你回复」上是唯一一定错的做法。 实测过一次:一条臂问了 4 个问题然后结束回合,30 秒后整个运行到此为止,PRD、issue 卡一个都没产出。
往下
python3 <SKILL_DIR>/../loop-it/scripts/loop_state.py route --scope <scope>(<SKILL_DIR> 是加载本技能时报告的 Base directory)
它按磁盘事实(documents/、issues/、检查点、工作树的改动)算出链上走到哪一段,next:
那一行就是你要用 skill 工具加载的技能名。照它说的加载——「我判断这整件事装得进一个上下文」
这句话属于被加载的那个技能,不属于这里。
链上的位置不靠回忆:那句话在你读完本技能二十多次调用之后就不在手边了,而 route 随时可查。
它还会拦越序——规划阶段却已经有实现改动时直接报错退出(踩过一次:PRD 落盘后直接开写代码,
整条链一步没走,直到用户当面追问才回头)。
用户后来的反馈一律当修订,不用回头重走。
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.

