What problem does it solve? It turns ad-hoc corrections and repeated feedback given to an AI coding agent into persistent, reusable instruction files so conventions are enforced automatically in future sessions. ## Core Features & Use Cases - Conversation Mining: Reviews the current conversation to extract corrections, preferences, and patterns the user has enforced (e.g., "always use X", "never do Y"). - Guided Clarification: Asks targeted questions about scope, affected file types, and whether a rule is mandatory or a preference when intent is unclear. - Iterative Refinement: Drafts the instruction file, identifies ambiguous parts, and refines them with the user before finalizing. - Use Case: After repeatedly correcting the agent to use named exports instead of default exports, invoke this Skill to save that rule as a persistent .instructions.md file applied automatically going forward. ## Quick Start Ask the agent to create an instructions file that enforces the coding convention you have been correcting it on during this conversation.