What problem does it solve? A single source often sounds confident but is wrong: aggregators return precise-looking numbers that differ from each other, and you never see the discrepancy until you open both. This Skill merges results from several sources (aggregators, government registries, web, exports) into one answer with deduplication, confidence scoring, and explicit display of disagreements instead of silently picking one version. ## Core Features & Use Cases - Confidence scoring: rates every fact as freshness x authority x agreement, using a tiered authority scale for Russian sources (EGRUL/FNS, GIR BO, kad.arbitr, FSSP at the top; aggregators like checko, rusprofile, Saby in the middle; forums at the bottom). - Deduplication and conflict handling: merges duplicates by completeness > authority > freshness, and shows divergent values as an explicit range with likely cause rather than choosing one number. - Attribution with tier markers: every claim is tagged with source, date, and a tier marker (confirmed / partial / unverified), plus a "What was NOT checked" footer. - Use Case: Ask how many lawsuits a counterparty has; the Skill collects counts from 4 aggregators, finds they range from ~500 to ~1500 due to different counting methodologies, and reports the spread while confirming FSSP debt and revenue figures that all sources agree on. ## Quick Start Ask the assistant to verify a company or fact through several sources and show where the sources agree and where they disagree.
