What problem does it solve? Investment research pipelines often pick data sources ad hoc, mixing low-trust vendor snapshots with official filings and leaving gaps unrecorded. This Skill reads an evidence plan and produces a deterministic source-call-plan that maps every required evidence type to capability-matched sources, trust tiers, fallbacks, and validation checks before analyst fan-out begins. ## Core Features & Use Cases - Capability-Based Source Selection: Selects sources by required evidence types and source capability rather than fixed use-case labels, ranking them by trust tier from T0 company/regulator disclosures down to T3 discovery sources. - Runtime Availability Probing: Merges live probe results into the plan, excludes runtime-unavailable sources from primary routing, and records missing tools or endpoints as unresolved data gaps. - Coverage Self-Check: Emits a machine-readable coverage block verifying every evidence type has a primary source and at least one fallback before the plan is declared complete. - Use Case: Before running analyst fan-out on a listed company, route financial statement evidence to SEC EDGAR or DART/KRX as primary with yfinance or FMP as fallback, and log any unavailable endpoints in unresolved-data-gaps.md. ## Quick Start Read the evidence plan in the active workspace and generate the source-call-plan, source-validation, and unresolved-data-gaps files under the 00_evidence directory.
