What problem does it solve? Teams repeatedly solve the same problems because lessons from completed work are never captured, or worse, a single anecdote gets promoted into an unjustified universal rule. This Skill converts completed evidence-backed work into governed reusable capability while preventing overfitting and doctrine sprawl. ## Core Features & Use Cases - Lesson Extraction & Classification: Identify the actual lesson from a milestone or incident and classify its generalizability as LOCAL ONLY, PATTERN, GENERALIZABLE, or UNKNOWN. - Smallest Durable Output Selection: Choose the minimal justified intervention, from no action or a documentation update to a regression test, policy candidate, skill candidate, or deprecation candidate. - Governed Activation Lifecycle: Route every candidate through DISCOVERED, CANDIDATE, TESTED, APPROVED, ACTIVE, and DEPRECATED/RETIRED stages with falsifiable acceptance tests and preserved negative learning. - Use Case: After resolving an incident where a scan executed and persisted results but the UI showed "No results", use this Skill to extract the generalizable lesson (separate execution, persistence, reporting, and presentation layers before repairing), create a regression test candidate, and route it through governance instead of rewriting the scanner. ## Quick Start Use the signalproof-learn skill to extract the reusable lesson from this completed milestone and recommend the smallest durable governed output.
