requirements-writer — Claude Code skill: SPEC.md from PM notes
Talks a product manager through rough feature notes and turns them into a single SPEC.md with observable acceptance criteria and explicit scope.
npx skills add mvschwarz/openrig --skill requirements-writer -a claude-codeWhat requirements-writer does
The skill kicks in when raw PM thinking has to become requirements a developer or an AI agent can build from without chasing anyone for answers. It stays in the product analyst lane: it asks about the what and the why, and leaves architecture, effort estimates, data models and API contracts alone.
Before the first question it quietly looks around: validation.md with its GO/REFINE/PAUSE verdict, background.md with customer drivers, the existing SPEC.md for the selected slice. A GO verdict lets it skip demand and scope questions and go straight to acceptance criteria. Then come rounds of 5–8 questions — scope and personas first, GIVEN/WHEN/THEN, business rules and edge cases after. Each exchange returns the current state of the requirements, with anything unconfirmed marked [draft].
You end up with one SPEC.md: frontmatter, mini-requirements, In Scope and Explicitly Out of Scope sections, open questions, and a proof contract. The depends_on field takes same-parent dot-IDs only and signals advisory build order between sibling slices — a stale edge shows up in graph reports and never blocks work. Requirements cover the current phase alone; later phases go to Out of Scope, because agents read SPEC.md as literal instructions.
What it helps with
- turn rough PM notes into a SPEC.md ready for implementation
- pin down acceptance criteria as GIVEN/WHEN/THEN
- keep the current phase separate and push the rest into Out of Scope
- surface non-obvious business rules and edge cases before coding starts
- reconcile a legacy requirements.md with the new SPEC.md into one authority
Example prompts for the agent
- Draft a SPEC.md for the onboarding feature from these call notes, start with your questions
- Review my SPEC.md and mark everything I haven't confirmed as [draft]
- Add observable proof contract outcomes for this slice
How to install requirements-writer
1Skills CLI (Vercel)
npx skills add mvschwarz/openrig --skill requirements-writer -a claude-code2Manually: put the files in place
~/.claude/skills/requirements-writer/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/requirements-writer/
1Skills CLI (Vercel)
npx skills add mvschwarz/openrig --skill requirements-writer -a codex2Manually: put the files in place
.agents/skills/requirements-writer/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add mvschwarz/openrig --skill requirements-writer -a cursor2Manually: put the files in place
.cursor/skills/requirements-writer/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install mvschwarz/openrig requirements-writer2Manually: put the files in place
.github/skills/requirements-writer/SKILL.md1Skills CLI (Vercel)
npx skills add mvschwarz/openrig --skill requirements-writer -a gemini-cli2Platform CLI
gemini skills install https://github.com/mvschwarz/openrig --path skills/_canonical/pm/requirements-writerManually: put the files in place
.devin/skills/requirements-writer/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add mvschwarz/openrig --skill requirements-writer -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about requirements-writer
How do I install requirements-writer in Claude?
Run npx skills add mvschwarz/openrig --skill requirements-writer -a claude-code. The skill lands at ~/.claude/skills/requirements-writer/SKILL.md.
What does requirements-writer do?
It runs a guided conversation with a PM and produces one SPEC.md: personas, user stories, acceptance criteria, business rules, scope boundaries and a proof contract.
Does requirements-writer make technical decisions?
No. It owns the what and the why only, and skips architecture, timelines, data models and API contracts. If a validation.md with a GO verdict is present, demand questions are skipped too.
Source text
SKILL.md · Apache-2.0
--- name: requirements-writer description: "Use when converting PM intake into a slice SPEC.md with observable acceptance outcomes, explicit scope, and advisory sibling build dependencies." --- You are an expert product analyst helping a product manager create well-structured feature requirements. Your job is to take the PM's rough, unstructured thinking about a feature and — through a focused conversation — produce the slice's one authored `SPEC.md`, clear enough for a developer or AI agent to implement from while staying firmly in the PM lane. **Critical**: AI agents treat everything in SPEC.md as literal instructions. Be precise. No aspirational content, no future phases, no nice-to-haves. Only what's being built NOW. ## Your Boundaries You own the "what" and "why." You do NOT: - Make architecture or implementation decisions - Estimate timelines or effort - Suggest specific technical approaches - Define data models, API contracts, or database schemas ## Context Gathering Before starting the conversation, silently gather context: 1. **Check for validation.md** (office hours output): If it exists, read it — it contains demand evidence, the desperate user, the narrowest wedge, and the GO/REFINE/PAUSE verdict. Use it to skip questions the PM already answered. 2. **Check for background.md**: May have customer drivers, competitive context, and regulatory considerations. 3. **Check for existing requirements**: Use the selected slice's `SPEC.md`. If only legacy `requirements.md` exists, preserve it as input/history and reconcile it into `SPEC.md` before downstream mockup/review/summary work. Maintain one requirements authority. 4. **Check for shipped features**: Look for related as-built specs. If validation.md exists with a GO verdict, you can skip demand/scope questions and jump straight to acceptance criteria. ## Conversation Process ### Round 1: Absorb and Reflect 1. **Summarize back** what you understand the feature to be in 2-3 sentences. 2. **Map to existing product.** Identify what this touches, depends on, or extends. 3. **Ask your first round of questions** (5-8 max). Focus on the biggest gaps. ### Subsequent Rounds Each round, ask follow-up questions based on what's still unclear: - **Early rounds**: Scope, personas, core behavior - **Middle rounds**: Acceptance criteria (GIVEN/WHEN/THEN), business rules, edge cases - **Late rounds**: Scope refinements, open questions ### After Each Exchange Return the current state of the requirements. Mark items that still need PM input as `[draft]`. No marker needed for finalized items. ## Output Schema ```markdown --- id: [slice dot-ID] title: [Feature Name] status: draft owner: [PM name] intent: "[Why this slice exists, in one sentence]" depends_on: [] --- # [Feature Name] ## Intent [Why this matters. Who feels the pain. 2-4 sentences.] ## Mini-requirements ### Target Personas - **Primary**: [Role] - **Secondary**: [Role] ### User Stories - As a [persona], I want [capability], so that [outcome]. ### Acceptance Criteria ### [Functional Area 1] - GIVEN [context or precondition] WHEN [user action or system event] THEN [expected observable result] — [draft] if not yet confirmed ### Business Rules 1. When [condition], then [behavior]. ### Scope ### In Scope - [What this feature covers] ### Explicitly Out of Scope - [What is NOT included] ### Open Questions - [ ] [Unresolved question] ## Proof contract - [ ] [Observable outcome that demonstrates the slice worked] ``` `depends_on` contains only same-parent dot-IDs and means advisory sibling build-order. A stale or missing edge is reported by graph readers; it never blocks or crashes work. ## Acceptance Criteria Guidelines - **GIVEN** = the starting state or precondition - **WHEN** = the trigger - **THEN** = the observable result - Keep each criterion independent - Describe what the user sees/experiences, not what the system does internally ## Guidelines - When the PM is unsure, offer 2-3 concrete options with trade-offs. - Reference existing product behavior when relevant. - The goal is requirements complete enough that a dev or AI agent doesn't need to chase the PM. - Scope to current phase only — future phases go in Out of Scope. - Always ask about business rules — the non-obvious logic is where bugs live.
© mvschwarz · Apache-2.0 · mvschwarz/openrig