To PRD Spec

This skill takes the current conversation context and codebase understanding and produces an agent-ready PRD/spec. Do NOT interview the user by default — synthesize what you already know.

If grill-me or grill-with-docs already sharpened the context, use that output. This skill turns available context into the artifact; it does not replace grilling when the user still wants that pressure step.

The artifact location should follow the repo convention or AFK artifact conventions. If no convention exists, write to docs/<task-slug>/<task-slug>.prd-spec.md. If issue tracker context exists and the user expects tracker publication, publish or prepare tracker-ready markdown instead.

Process

1. Explore the repo to understand the current state of the codebase, if you haven't already. Use the project's domain glossary vocabulary throughout the PRD/spec, and respect any ADRs in the area you're touching.

2. Sketch out the seams at which you're going to test the feature. Existing seams should be preferred to new ones. Use the highest meaningful seam possible. The fewer seams across the codebase, the better; one honest high-level seam is ideal when it covers the behavior.

Check with the user that these seams match their expectations.

3. Write the PRD/spec using the template below. If a PM PRD already exists, preserve its product intent and strengthen only what is missing for agent execution: behavior, acceptance criteria, implementation decisions, testing seams, or out-of-scope boundaries.

4. After writing a local PRD/spec, run plannotator annotate --gate <path-to-prd-spec> when Plannotator is available. Treat returned annotations as requested changes and update the artifact before handing it to slicing or execution. If Plannotator is unavailable, say the artifact path and continue.