What problem does it solve? AI-generated research reports often contain claims, quotes, and citations that cannot be traced back to their sources. This Skill truth-checks every statement in a report against the original source files, correcting errors in place and documenting failures instead of hiding them. ## Core Features & Use Cases - Claim-by-claim verification: Locates each claim's source file, compares quotes against the original text, and assigns a status of verified, corrected, unsupported, contradicted, or unresolved. - In-place correction and status updates: Fixes minor errors directly in the report, updates the artifact status (pass, pass_with_corrections, partial, fail, blocked), and syncs the Navigation Dashboard status line. - Citation integrity checks: Confirms every cited source path exists in the raw document collection and refuses to certify claims that cannot be traced to a registered source. - Use Case: After an AI agent drafts a report summarizing findings across dozens of interview transcripts, run this Skill to confirm every quote matches the transcripts, fix misattributions, and flag any unsupported claims before the report is shared. ## Quick Start Verify the draft report in agent_reports against the source files in raw and correct any unsupported claims in place.
