What problem does it solve? Web research often produces unverified, outdated, or marketing-driven claims presented as facts. This Skill enforces a rigorous pipeline where every load-bearing claim must survive adversarial verification before appearing in the final report, eliminating the primary failure mode of trusting unverified web content. ## Core Features & Use Cases - Structured Research Pipeline: Decomposes a question into 3-6 complementary search angles, fans out web searches, fetches up to 15 deduplicated sources, and extracts falsifiable claims with supporting quotes. - Adversarial Verification: Runs 3 independent skeptical checks per claim on the top 25 claims, killing any claim with 2 or more refutations and tracking the vote tally. - Confidence-Ranked Reporting: Produces a Markdown report with an executive summary, findings rated high/medium/low confidence, a refuted-claims section for transparency, caveats, open questions, and a source quality table. - Use Case: An analyst performing due diligence on a vendor asks for a fact-checked report; the Skill searches multiple angles, verifies each vendor claim against independent sources, and delivers a cited report distinguishing what the product does from what the vendor claims. ## Quick Start Ask the agent to do deep research on your question, for example: run deep research on whether solid-state batteries will reach mass-market EVs by 2030 and give me a fact-checked report.
