What problem does it solve?
Early-stage ideas are fragile: searching only the user's first wording misses better framings, jumping to enthusiasm creates roadmap debt, and listing prior art alone never answers what to do next. This Skill gives an agent a disciplined FRAME → DIVERGE → RESEARCH → CROSS-POLLINATE → STRESS-TEST → SYNTHESIZE → DECIDE workflow so fuzzy ideas are tested against evidence before anyone commits to building.
Core Features & Use Cases
- Divergent framing and prior-art mapping: Generates 6-10 alternative angles, then maps existing solutions across local code, GitHub repositories, packages, and web sources with strong/moderate/weak confidence markers.
- Multi-engine web research: Consolidates results from Tavily, Serper, and Exa (with DuckDuckGo fallback), deduplicating by canonical URL and cross-confirming claims across engines.
- Perspective stress-testing and verdicts: Runs Critical Architect, Visionary Entrepreneur, and Product lenses, concedes weak evidence, and outputs a decision such as Build RFC, Prototype First, Narrow, Park, or Do Not Build.
- Use Case: A user asks "Is it worth building a CLI that summarizes GitHub PR discussions?" The agent reframes the idea, checks existing tools and package momentum, stress-tests the thesis, and delivers a decision brief with cited sources and one concrete next step.
Quick Start
Ask the agent to validate whether your idea is worth building, requesting prior-art research across GitHub, packages, and the web with a final build-or-park recommendation.