What problem does it solve? Valuable lessons learned during debugging, feature implementation, and design discussions are often lost between sessions. This Skill captures those insights as executable code-spec documents so future developers and AI sessions can implement safely without repeating mistakes. ## Core Features & Use Cases - Structured Spec Updates: Classifies learnings into design decisions, conventions, patterns, anti-patterns, common mistakes, and gotchas, then writes them into the correct spec file under .trellis/spec/. - Executable Contract Enforcement: Requires signatures, request/response contracts, validation and error matrices, Good/Base/Bad cases, and test assertion points for infra or cross-layer changes. - Code-Spec vs Guide Separation: Routes "how to implement" content into backend/ or frontend/ specs and "what to consider" checklists into guides/. - Use Case: After fixing a subtle Windows stdout encoding bug, run this Skill to add a Wrong vs Correct example and a convention entry to backend/script-conventions.md so the mistake is never repeated. ## Quick Start Ask the AI to update the relevant code-spec with what was just learned from the completed task or bug fix.
