What problem does it solve? Valuation and corporate-finance analyses fail when inputs are invented, unsourced, or silently computed. This Skill collects every analysis input — financial statements, footnotes, market data, macro rates, reference tables, and peer sets — and records exactly where each figure came from, which fallbacks were applied, and what remains missing. ## Core Features & Use Cases - Provenance-first collection: Every field in the output JSON carries source, as-of date, retrieval date, and a document reference precise enough to find the page again. - Structured gap accounting: Missing data is never invented; each gap records its status, the fallback ladder rung applied, its cost, and which downstream stages it constrains. - Validation gates: Runs six statement reconciliation ties and a reference-table vintage check (R4 rule) via Python scripts, and assesses the G1 gate predicate before handing off. - Use Case: As the first stage of a valuation orchestrator run, collect ten years of statements, footnotes, market data, Damodaran reference snapshots, and two peer sets for a public company, then emit raw-financials.json, market-data.json, sources.md, and gaps.json for downstream DCF and relative-valuation stages. ## Quick Start Ask the valuation orchestrator to run the data-collection stage for a company by supplying a mandate.json with mode, company identity, valuation currency, and valuation date.
