What problem does it solve? Researching multiple subjects at once often produces unverified claims, fabricated quotes, and inconsistent standards that reach published deliverables. This Skill coordinates a wave of parallel research agents, independent fact-checkers, and corrections agents so every claim is verified before anyone acts on it. ## Core Features & Use Cases - Parallel research passes: Launch one agent per subject against an identical schema, with per-item write isolation and compact returns that protect the orchestrator's context. - Independent fact-checking: Each result is checked by an agent that did none of the original research, with named failure modes like fabricated quotes, dead sources, and over-corrections. - Ruled corrections and cross-subject sweeps: The orchestrator rules on every finding, a corrections agent applies rulings while re-deriving numbers, and final sweeps catch consistency, formatting, and stale-count defects. - Use Case: Run a vendor comparison across five products where the results will be published: define the schema once, research all five in parallel, fact-check each independently, and synthesize only from corrected records. ## Quick Start Use the research-wave skill to research these five vendors in parallel against a shared schema and independently fact-check every claim before summarizing.
