What problem does it solve? Answering questions that require current, source-backed information is slow and unreliable when done manually, and a single search method often misses depth or accuracy. This Skill combines Claude WebSearch with Perplexity's sonar-deep-research model behind an automatic 2-tier router, so simple fact checks stay fast and free while complex analyses get deep, cross-checked research with citations and confidence ratings. ## Core Features & Use Cases - Automatic 2-Tier Routing: QUICK tier uses Claude WebSearch for facts, prices, and short answers; DEEP tier uses Perplexity sonar-deep-research via OpenRouter plus a parallel WebSearch cross-check for comparisons, market analysis, and multi-aspect questions. - Structured, Source-Backed Answers: Every response includes a summary, aspect-structured details, sources separated by origin (WebSearch vs OpenRouter/Perplexity), a confidence rating, and the tier used. - Composable Service Design: Returns raw findings plus synthesis without overwriting the calling context, so other skills like ideation, implement, or sprint-review can consume results directly. - Use Case: Ask "How does Clerk vs Auth0 compare for a Next.js SaaS in 2025?" and receive a deep-research comparison with pricing, architecture trade-offs, cited sources, and a confidence rating in under two minutes. ## Quick Start Set your OPENROUTER_API_KEY in the .env file, then ask the assistant to research a topic such as "research the current state of local-first sync engines" and receive a structured, cited answer.
