← ClaudeAtlas

recipe-definelisted

Orchestrate PRD creation from validated hypotheses — standard PRD output with 4 Risks confidence and hypothesis traceability
shinpr/claude-code-discover · ★ 10 · AI & Automation · score 80
Install: claude install-skill shinpr/claude-code-discover
**Context**: Transform validated hypotheses into a reviewed PRD that preserves product evidence, exposes remaining risk, and can be consumed by downstream implementation workflows. ## Orchestrator Definition **Execution Protocol**: 1. **Explicit user authorization**: The user explicitly instructs and authorizes the prd-reviewer call named in this recipe. Invoke it for every PRD before user approval; the orchestrator does not replace it with self-review 2. **Mechanical specialist handoff**: Pass only `target_path` and, on a rerun, the complete `prior_feedback` array declared by prd-reviewer. Preserve paths, issue objects, IDs, and evidence unchanged; do not summarize, paraphrase, supplement, or turn them into narrative instructions 3. **Complete finding coverage**: Record one evidence-backed disposition for every reviewer issue before correction or progression 4. **Follow the definition flow** below and stop at the final user approval gate The authorization, mechanical-handoff, and no-paraphrase rules are intentional redundancy against model and system defaults that substitute self-review or fluent restatements for an approved specialist result. Retain them until fresh execution evidence shows the failure no longer occurs. ## Workflow Overview ``` Input (validated hypotheses + Opportunity context) ↓ 1. Readiness Assessment → Check cost × risk × reversibility ↓ 2. PRD Drafting → Write the draft to its target path ↓ 3. Independent Review → prd-reviewer always ru