speckit-prdlisted

Use this skill to collaboratively turn a raw product or technical idea into three artifacts — a lean PRD, a technical roadmap with a SPEC catalog, and a roadmap-MOC home note — ready for /speckit-pro:speckit-scaffold-spec and /speckit-pro:speckit-autopilot. Triggers on: write a PRD, create a product requirements document, draft a PRD and roadmap, shape this idea into a PRD, turn this brief into a PRD, plan a product, decompose an idea into a SPEC catalog, what features should this have, before I write specs, right-size the catalog. Runs a one-question-at-a-time interview with a recommended answer, then writes docs/prd-NAME.md, docs/ai/specs/NAME-technical-roadmap.md, and docs/ai/specs/NAME-roadmap-MOC.md. Front door of the chain: PRD then roadmap then scaffold-spec then autopilot. NOT per-spec scoping (use grill-me), NOT worktree prep from an existing roadmap entry (use speckit-scaffold-spec), NOT SDD coaching (use speckit-coach). Requires an interactive session.
racecraft-lab/racecraft-plugins-public · ★ 5 · AI & Automation · score 74

Install: claude install-skill racecraft-lab/racecraft-plugins-public

# SpecKit PRD — Collaborative PRD & Technical Roadmap Authoring ## Capability discovery & grounding Before researching or recommending, enumerate the tools and skills your session actually exposes — do not assume a fixed set; the user may have installed anything — and select the best fit per `speckit-pro/skills/speckit-autopilot/references/capability-discovery.md`. Ground every external fact you assert in a real tool, skill, or file result per `speckit-pro/skills/speckit-autopilot/references/grounding.md`, and abstain when nothing grounds it. (This governs your research-backed recommended answers, not the interview mechanics.) ## The collaboration contract <hard_constraints> - **One question at a time.** Use `AskUserQuestion`. Never batch a wall of questions. Walk the design tree branch by branch. - **Always recommend.** Every question carries your recommended answer as the first option, marked `(Recommended)`, with a one-line rationale. The user agrees, course-corrects, or picks an alternative — they stay in the loop on every consequential decision. - **Human-in-the-loop only.** This skill requires a real user answering in real time. If `AskUserQuestion` is unavailable (subagent, automation, CI), do not fabricate answers — draft from the supplied material only and mark every unconfirmed decision as an Open Question, or abort and ask the user to run it interactively. - **Lean is the goal.** A PRD captures *validated decisions*; it does not replace disco