Idea to PRD Workflow

Converts a raw idea into a production-ready PRD by running three skills in sequence. Each step produces a document that the next step reads before proceeding. Do not skip steps — skipping validation leads to building the wrong thing; skipping design leads to an incoherent PRD.

Skill Chain

idea-validator
    ↓ [Gate 1: Quick Verdict ≠ "Skip it"]
brainstorming
    ↓ [Gate 2: design doc exists in docs/designs/]
prd-writer
    ↓ [Gate 3: PRD sections complete]
→ Hand off to planning or full-stack-build

---

Step 1 — Validate the Idea

Load skill: idea-validator

Run a rapid validation of the idea before investing any design or engineering time.

Input: One sentence describing the idea, the target user, and the core problem it solves.

Questions idea-validator answers:

• Is there demand evidence or only assumed demand?
• Is market saturation already high?
• Can a solo builder ship an MVP in 2-4 weeks?
• Is there a clear monetization path?

Output: Quick Verdict — one of:

• Build it — validated, proceed to Step 2
• Maybe — proceed to Step 2 with caveats documented
• Skip it — stop here; document reason in docs/ideas/YYYY-MM-DD-[idea]-skip.md

Gate 1: If Quick Verdict is "Skip it", stop the workflow. Record the verdict and reason. Revisit in 30 days or with new evidence. Do not proceed to Step 2 out of attachment to the idea.

---

Step 2 — Design the Solution

Load skill: brainstorming

With the idea validated, design the solution before specifying it.

Input: The idea + Quick Verdict + any caveats from Step 1.

Questions brainstorming answers:

• What exactly does the MVP do? (scope boundary)
• What are the 2-3 viable approaches?
• What are the trade-offs between approaches?
• What's the data model and key flows?
• What are the error cases and edge cases?

Brainstorming protocol:

1. Load context (similar features in the codebase if applicable)
2. Ask 1 Socratic question at a time
3. Propose 2-3 approaches with explicit trade-offs
4. Verify each design section before moving on
5. Save the approved design to docs/designs/YYYY-MM-DD-[topic].md

Gate 2: docs/designs/YYYY-MM-DD-[topic].md must exist and be reviewed before opening the PRD. The design doc is the source of truth for the PRD — the PRD translates design decisions into stakeholder language.

---

Step 3 — Write the PRD

Load skill: prd-writer

With a validated idea and an approved design, write the PRD.

Input: The design doc from Step 2 + target stage (planning/kickoff/solution review).

PRD sections that must be complete before the gate:

• Problem Statement + target user
• Success Metrics (at least 1 primary metric with target value)
• Solution Overview (aligned with design doc)
• Scope: In / Out of scope for this iteration
• Rollout plan (percentage or feature flag)

Output: docs/prd/YYYY-MM-DD-[feature].md

Gate 3: PRD with all sections above complete. Incomplete PRDs (missing metrics or scope) lead to scope creep and unclear success criteria. Do not hand off to engineering until metrics are defined.

---

Handoff

After Gate 3:

• Next workflow: full-stack-build (if building now) or planning (for more detailed task breakdown)
• Archive: Link the design doc and PRD in the PRD's "References" section

---

Output Artifacts

docs/ideas/YYYY-MM-DD-[idea]-skip.md   (only if Gate 1 = Skip it)
docs/designs/YYYY-MM-DD-[topic].md     (Step 2 output)
docs/prd/YYYY-MM-DD-[feature].md       (Step 3 output)

Gates Summary

Gate	Condition to pass
After Step 1	Quick Verdict = "Build it" or "Maybe"
After Step 2	Design doc saved and reviewed
After Step 3	PRD has problem, metrics, solution, scope, rollout

Common Antipatterns

• Skipping validation because you're excited — Excitement is not market demand. Run Step 1 even if you're sure.
• Writing the PRD before designing — PRDs that skip design discovery are wish lists, not specs. Design first.
• "Maybe" verdicts that never get revisited — A "Maybe" with no follow-up becomes a "Build it" by default. Document the conditions under which "Maybe" becomes "Build it".

See Also

• idea-validator — For standalone idea validation outside this workflow
• brainstorming — For design exploration outside this workflow
• prd-writer — For PRD writing outside this workflow (e.g., updating an existing PRD)
• full-stack-build — The workflow that follows this one (build phase)