What problem does it solve? Coordinating multi-agent OSINT research is error-prone: investigators produce malformed findings, fact-checkers cite unverified sources, and loops run without stopping criteria. This Skill orchestrates bounded execution cycles that spawn investigator and fact-checker agents, validate their outputs, and stop at editorial gates. ## Core Features & Use Cases - Bounded execution cycles: Runs up to 5 autonomous cycles, each spawning an investigator agent, validating findings.json, then spawning a fact-checker agent with structural and evidence validators. - Correction and stall handling: Re-spawns agents for structural, evidence, or editorial corrections with hard caps, and triggers a Stall Protocol when cycles are exhausted. - Readiness gating: Evaluates six readiness criteria (source independence, unresolved disputes, document trail, gap assessment) before resolving to the Gate 1 human review. - Use Case: An investigator agent gathers findings on a lead, the fact-checker independently verifies each claim against archived sources, and the cycle repeats until readiness criteria pass or the case is blocked. ## Quick Start Invoke this phase after methodology approval to run autonomous investigation cycles against the case directory until findings meet readiness criteria.
