What problem does it solve? Investment research requests often require evidence beyond company filings, such as market context, trade data, regulatory signals, and macro indicators. This Skill structures those open-ended requests into a formal question decomposition and evidence plan before any analyst fan-out begins, so downstream analysts know exactly which evidence types, source capabilities, and validation gates to apply. ## Core Features & Use Cases - Question Decomposition: Breaks a raw user request into entities, open-ended subjects, geographies, time horizons, claim types, and required evidence types, preserving the user's surface terms. - Evidence Planning: Maps each evidence type to source capability needs, signal primitives, and validation gates, including a mandatory narrative plausibility gate based on Damodaran's possibility, plausibility, and probability filters. - Role Slice Emission: Appends a machine-readable YAML block assigning required evidence types to six analyst roles (financial, fundamental, valuation, technical, macro-sentiment, risk-scenario). - Use Case: When a user asks to analyze a company with a tariff catalyst, run this Skill first to produce question-decomposition.md and evidence-plan.md in the workspace's 00_evidence folder, which downstream analyst skills then consume. ## Quick Start Ask the agent to run evidence planning on the current investment research request so it writes the question decomposition and evidence plan into the workspace evidence folder.
