What problem does it solve? Investigating complex topics often produces shallow summaries built on unverified claims, with no way to trace findings back to sources or resume an interrupted research run. This Skill enforces systematic multi-source fact-finding where every load-bearing claim must survive adversarial verification before entering the report. ## Core Features & Use Cases - Parallel Source Analysis: Dispatches subagents to analyze batches of up to 10 sources, extracting falsifiable claims with supporting quotes and credibility classes, tracked in a resumable source-tracker.md. - Iterative Discovery: Follows source → insight → new source chains up to 3 levels deep, detecting duplicates and circular references to keep the run convergent. - Adversarial Claim Verification: Runs a three-vote skeptical panel per claim, classifying each as confirmed, refuted, or unverified, so weak or outdated claims never become findings. - Use Case: Ask it to research the current state of a technology or market; it produces a cited report with confidence ratings, debates, refuted-claim transparency, and explicit research gaps. ## Quick Start Ask the agent to run deep research on your topic, for example: research the current state of solid-state battery commercialization with a focus on manufacturing scalability.
