brainstorm

Codex compatibility note:

• Invoke $skill-name: load the registered .agents/skills/<name>/SKILL.md and execute it with Codex tools; no separate Skill tool or Claude mode switch is needed. Author .claude/**; mirrors are generated.
• Track steps before execution and keep progress/evidence current. Explicit skill requests execute their protocol. Use the native question tool for required questions.
• Map foreign tool names to available host capabilities. A name mismatch needs no approval; an actually unavailable required capability stops with its step/evidence and a user question before adaptation.
• Subagent authorization: when a skill is user-invoked or AI-detected and its protocol requires subagents, that skill activation authorizes use of the required spawn_agent subagent(s) for that task.
• Preserve protocol order unless the user approves a deviation. Workflows use $start-workflow: gates stay fixed; other steps may flex with logged reasons and evidence.

[BLOCKING] Execute skill steps in declared order. NEVER skip, reorder, or merge steps without explicit user approval. Inside a workflow run, $start-workflow → Workflow-step mode bounds this rule. [BLOCKING] Before each step or sub-skill call, update todo tracking: set in_progress when step starts, set completed when step ends. [BLOCKING] Every completed/skipped step MUST include brief evidence or explicit skip reason. [BLOCKING] If Task tools are unavailable, create and maintain an equivalent step-by-step plan tracker with the same status transitions.

Quick Summary

Goal: Facilitate evidence-backed PO/BA Double-Diamond ideation that separates problem discovery from solution evaluation and delivers either a validated, ranked 3–5-candidate shortlist with problem/value hypotheses, each riskiest assumption, and cheapest validation test plus one recommendation—or, in Multi-Opportunity Discovery mode, a ranked 3–8-item RICE map for user selection—so the team commits to the right problem and solution, never a flat idea list.

Summary:

• Ordered core: P0 Setup (scenario/role/known + context) → P1 Problem Framing/diverge (POV, 5 Whys/Fishbone, JTBD, HMW) → P2 Opportunity Framing/converge (OST, Lean Canvas, ERRC, Value Proposition) → P3 Ideation/diverge (SCAMPER, Crazy 8s, Brainwriting, Impact Mapping, Analogy; 25–40 ideas) → P4 Evaluation/converge (Dot Vote, RICE, Kano, 2×2, MoSCoW; shortlist 3–5) → P5 Validation (problem/value cards, RAT, cheapest test, Build-Measure-Learn) → P6 Decision (one recommendation) → P7 Documentation/Handoff.
• Purpose and gates: Run ask user question tool in P0 first; separate diverge from converge; test every top-3 candidate before build; Multi-Opportunity Discovery ranks 3–8 opportunities and uses multi-select.
• Routing: Resolve --mode=roadmap|scope before P0. Roadmap hands outcome/milestone framing to $product-roadmap; scope amends one approved plans/{plan-id}/scope-brief.md and stops before scenario/plan work (plans root default plans/; docsRoots.plans.path in docs/project-config.json overrides it).
• Handoff boundary: Apply isLargeIdea; when true, carry one complete large_idea_decomposition block in the owning handoff. Default mode offers user-selected handoff ($initiative, $work-item --mode=refine, $plan, etc.); Multi-Opportunity Discovery hands selected items to the per-opportunity task loop.

Workflow:

1. Session Setup — ask user question tool detects scenario, role, known context; load domain or market context.
2. Problem Framing/diverge — POV, root cause when applicable, JTBD, HMW.
3. Opportunity Framing/converge — scenario-specific OST, Lean Canvas, ERRC, Value Proposition.
4. Ideation/diverge — SCAMPER, Crazy 8s, Brainwriting, Impact Mapping, Analogy; no judgment.
5. Evaluation/converge — Dot Vote, RICE, Kano, 2×2, MoSCoW; rank 3–5 candidates or a 3–8 opportunity map.
6. Hypothesis Validation — problem/value cards, RAT, cheapest test, Build-Measure-Learn for top 3.
7. Decision, Documentation & Handoff — recommend one option by default; document and route user-approved next steps.