Analysis updated 2026-05-18
Turn a vague AI agent or AI product idea into a structured PRD through guided clarifying questions.
Decide which AI capabilities, such as RAG, memory, or tool calling, a product idea actually needs before overdesigning it.
Convert existing meeting notes or a competitor analysis into a cleaner, gap-filled PRD.
Get a final review pass on a draft PRD for conflicts, overdesign, and MVP risk before sharing it with a team.
| huohuodong94-crypto/prd-copilot-skill | 3b1b/site_demo | 5bv57zcm44-max/noxus-ai-open-whatsapp | |
|---|---|---|---|
| Stars | 27 | 27 | 27 |
| Language | — | Shell | TypeScript |
| Last pushed | — | 2021-04-10 | — |
| Maintenance | — | Dormant | — |
| Setup difficulty | easy | easy | moderate |
| Complexity | 1/5 | 1/5 | 4/5 |
| Audience | pm founder | general | developer |
Figures from each repo's GitHub metadata at analysis time.
Copy the skill folder into your local Codex skills directory, no build step or external dependencies required.
PRD Copilot is a Skill for Codex, an AI coding assistant, that helps a product manager turn a vague product idea into a clear, concise product requirements document, commonly called a PRD. It is aimed especially at early stage ideas for AI products and AI agent products, and it exists to stop an assistant from rushing into writing a full PRD before the requirements are actually clear. When someone describes an idea, such as wanting to build an AI agent or an AI travel assistant, the skill first checks whether the request is actually a product requirements task at all. If it is, it checks for basic missing pieces like the product's goal, target users, core pain point, usage scenario, business goal, core features, and constraints. If key information is missing, it asks at most five focused clarifying questions rather than guessing. Once there is enough context, it builds out the product definition, covering positioning, user personas, pain points and value, business flow, user journey, and minimum viable product scope, referencing competitors only when useful. For AI or agent products specifically, it only selects AI capabilities that the scenario actually needs, choosing from options like prompt design, retrieval augmented generation, memory, tool calling, standardized tool access, multi step workflows, or multiple specialized agents working together. The final output is a single, concise markdown PRD file covering background, product goals, target users, core scenarios, feature requirements, user flow, page or interaction design, AI capability design, data and metrics, edge cases and risks, and an MVP and release plan. Before finishing, it reviews the draft for conflicts, overdesign, and MVP risk. Installing it means copying the skill folder into a local Codex skills directory and then invoking it by name in a new Codex session. The documentation is written mostly in Chinese, with some skill names and technical terms in English.
A Codex Skill that asks clarifying questions to turn a vague AI product idea into a clear, concise Markdown PRD, choosing only the AI capabilities the idea actually needs.
No license information is stated in the README.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly pm founder.
This repo across BitVibe Labs
Verify against the repo before relying on details.