continuous-learning-v2

Capturing session activity as reusable knowledge uses observation hooks to detect patterns in prompts and tool usage, forming atomic instincts with confidence scoring. This pipeline evolves validated patterns into skills, commands, or autonomous agents.

Atomic instincts are small, reusable units of knowledge derived from session activity, scored by confidence to guide their evolution. They apply to project-scoped learning by capturing specific patterns and determining when they should be promoted.

Promoting reusable instincts into commands or autonomous agents is managed through evolution rules with project and global scope management. This pipeline evolves high-confidence atomic instincts into skills, commands, or agents.

Getting started with hook-based observation requires enabling observation hooks to capture session activity. Running status checks then allows you to view learned atomic instincts and track their evolution into skills, commands, or agents.