What problem does it solve? When facing unfamiliar domains, new frameworks, or design decisions with multiple valid directions, relying on model memory alone risks outdated or incomplete answers. This Skill performs structured online research so decisions are grounded in what actually exists and how others solve the problem. ## Core Features & Use Cases - Adaptive research depth: Quick (1-2 rounds), Normal (2-4), or Deep (4-8) search rounds, with automatic escalation when findings contradict each other. - Dual research tools: WebSearch for broad exploration of unfamiliar territory and WebFetch for targeted reading of documentation, GitHub discussions, blog posts, and forum threads. - Three output modes: Absorb findings silently into the work, Discuss options ranked by end-result quality, or Challenge the current direction when research surfaces risks or anti-patterns. - Use Case: Before choosing a state management approach for a new app, trigger the skill to survey community consensus, compare trade-offs from real-world write-ups, and present the strongest options with sources. ## Quick Start Ask the agent to research how other teams handle a problem you are facing, for example by saying "research what approaches exist for this design decision before we commit".
