instinct-evolve

To convert learned AI patterns into permanent rules, you can use a learning pipeline evolution skill that reviews qualified instincts and proposes promoting them into durable artifacts like memory entries, rules, or new skills.

An instinct evolution phase in an AI learning pipeline is the process of reviewing high-confidence behavioral patterns and promoting them into structured knowledge artifacts such as MEMORY.md, CLAUDE.md, or new skill definitions.

To promote qualified instincts to structured knowledge, the system reads instinct data from .ai/instincts/instincts.yml, evaluates predefined criteria like confidence and evidence, and writes approved patterns to durable artifacts.

Yes, promoting AI instincts to durable artifacts requires explicit human approval, operating as a human-in-the-loop process to ensure user control before applying any changes to memory entries, rules, or skills.

Yes, you can filter AI instincts based on confidence, evidence, and status using predefined evolution criteria to ensure only high-quality patterns are proposed for artifact generation.

Durable artifacts generated from promoting AI instincts include MEMORY.md entries, CLAUDE.md rules, and entirely new skills, transforming recurring behavioral patterns into permanent, actionable knowledge.