What problem does it solve? Long-term investors struggle to tell whether a change in their research reports reflects real fundamental change or just rewording and price noise. This Skill compares two thesis snapshots and separates fact changes from wording changes, so you only revisit an investment when the evidence actually moved. ## Core Features & Use Cases - Evidence-Normalized Comparison: Extracts core assumptions, red lines, valuation anchors, management quality, and moat judgments from two reports into a single comparison table, judging each dimension as Improved, Unchanged, or Weakened. - Rigorous Numeric Verification: Routes all valuation math (PE, PB, FCF yield, market cap, three-scenario targets) through the bundled financial_rigor.py script using exact decimal arithmetic, with multi-source cross-validation. - Three Operating Modes: Compare two specified report paths, auto-discover dated snapshots in the reports directory, or gracefully handle a missing baseline by guiding you to build one first. - Use Case: After a new earnings season, run the drift check on your Tencent thesis snapshots to confirm whether margin compression is a real fundamental weakening or just a rephrased report, and get a clear action migration such as Hold to Reduce. ## Quick Start Ask the agent to run a thesis drift check on a company by providing the company name plus the paths of the old and new thesis reports, or just the company name to auto-compare snapshots in the reports folder.
