spec
Write or revise requirements with observable acceptance criteria. Triggers on "write spec", "写需求文档".
SKILL.mdskills/01-product/spec/SKILL.md
GitHubSpec
When to use
- Write or revise requirements with observable acceptance criteria.
Not for: architecture for technical design; documentation-audit for descriptive drift.
Steps
Read relevant material and confirmed decisions; resolve unknown goals, scope and acceptance without asking the user to guess inspectable facts.
Describe scenarios, behavior, rules and relevant exceptions; technical implementation belongs in TECH.
Connect requirements to observable acceptance cases. Cover permissions, repetition, failure and recovery where needed; preserve confirmed scope and stable IDs.
Prepare a readable draft at the approved path with open decisions and risk-based audit items; confirm governing requirements before dependent implementation.
Output: Suggested paths, subject to human audit: docs/PRD.md; optionally docs/vX.Y/prd.md for separately maintained release scope.
Verify
- Core requirements have observable acceptance cases; approval and unresolved scope are honestly reported.
References
engineering-principles.md — shared decision boundaries, scope and verification.
product-principles.md — product discipline (need≠feature, outcomes over outputs, say no to good ideas, the real competitor is the workaround).
references/readable-prd-guide.md — readable-spec craft rules: layout density, Mermaid figures, requirement cards, human audit & OPN formats, update discipline, engineering sections.
references/prd-patterns.md — PRD structure, user stories, Given/When/Then acceptance criteria, INVEST, Non-Goals, success-criteria reframing, anti-patterns.
references/prioritization.md — RICE, ICE, Kano, MoSCoW, Value×Feasibility matrix, true-need vs false-need filter.
references/metrics-frameworks.md — North Star metric, AARRR funnel, retention curves, cohort analysis, Hook Model, A/B testing discipline, data-driven loop.
Human audit — artifact consent and review.