You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

draft-and-review.md 4.0KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687
  1. **Language:** Use `{communication_language}` for all output.
  2. **Output Language:** Use `{document_output_language}` for documents.
  3. **Output Location:** `{planning_artifacts}`
  4. **Paths:** Bare paths (e.g. `agents/foo.md`) resolve from the skill root.
  5. # Stage 4: Draft & Review
  6. **Goal:** Produce the executive product brief and run it through multiple review lenses to catch blind spots before the user sees the final version.
  7. ## Step 1: Draft the Executive Brief
  8. Use the template at `{workflow.brief_template}` as a guide — adapt structure to fit the product's story.
  9. **Writing principles:**
  10. - **Executive audience** — persuasive, clear, concise. 1-2 pages.
  11. - **Lead with the problem** — make the reader feel the pain before presenting the solution
  12. - **Concrete over abstract** — specific examples, real scenarios, measurable outcomes
  13. - **Confident voice** — this is a pitch, not a hedge
  14. - Write in `{document_output_language}`
  15. **Create the output document at:** `{planning_artifacts}/product-brief-{project_name}.md`
  16. Include YAML frontmatter:
  17. ```yaml
  18. ---
  19. title: "Product Brief: {project_name}"
  20. status: "draft"
  21. created: "{timestamp}"
  22. updated: "{timestamp}"
  23. inputs: [list of input files used]
  24. ---
  25. ```
  26. ## Step 2: Fan Out Review Subagents
  27. Before showing the draft to the user, run it through multiple review lenses in parallel.
  28. **Launch in parallel:**
  29. 1. **Skeptic Reviewer** (`agents/skeptic-reviewer.md`) — "What's missing? What assumptions are untested? What could go wrong? Where is the brief vague or hand-wavy?"
  30. 2. **Opportunity Reviewer** (`agents/opportunity-reviewer.md`) — "What adjacent value propositions are being missed? What market angles or partnerships could strengthen this? What's underemphasized?"
  31. 3. **Contextual Reviewer** — You (the main agent) pick the most useful third lens based on THIS specific product. Choose the lens that addresses the SINGLE BIGGEST RISK that the skeptic and opportunity reviewers won't naturally catch. Examples:
  32. - For healthtech: "Regulatory and compliance risk reviewer"
  33. - For devtools: "Developer experience and adoption friction critic"
  34. - For marketplace: "Network effects and chicken-and-egg problem analyst"
  35. - For enterprise: "Procurement and organizational change management reviewer"
  36. - **When domain is unclear, default to:** "Go-to-market and launch risk reviewer" — examines distribution, pricing, and first-customer acquisition. Almost always valuable, frequently missed.
  37. Describe the lens, run the review yourself inline.
  38. ### Graceful Degradation
  39. If subagents are unavailable:
  40. - Perform all three review passes yourself, sequentially
  41. - Apply each lens deliberately — don't blend them into one generic review
  42. - The quality of review matters more than the parallelism
  43. ## Step 3: Integrate Review Insights
  44. After all reviews complete:
  45. 1. **Triage findings** — group by theme, remove duplicates
  46. 2. **Apply non-controversial improvements** directly to the draft (obvious gaps, unclear language, missing specifics)
  47. 3. **Flag substantive suggestions** that need user input (strategic choices, scope questions, market positioning decisions)
  48. ## Step 4: Present to User
  49. **Headless mode:** Skip to `prompts/finalize.md` — no user interaction. Save the improved draft directly.
  50. **Yolo and Guided modes:**
  51. Present the draft brief to the user. Then share the reviewer insights:
  52. "Here's your product brief draft. Before we finalize, my review panel surfaced some things worth considering:
  53. **[Grouped reviewer findings — only the substantive ones that need user input]**
  54. What do you think? Any changes you'd like to make?"
  55. Present reviewer findings with brief rationale, then offer: "Want me to dig into any of these, or are you ready to make your revisions?"
  56. **Iterate** as long as the user wants to refine. Use the "anything else, or are we happy with this?" soft gate.
  57. ## Stage Complete
  58. This stage is complete when: (a) the draft has been reviewed by all three lenses and improvements integrated, AND either (autonomous) save and route directly, or (guided/yolo) the user is satisfied. Route to `prompts/finalize.md`.