agent-memory-hygiene

Agent memory refactoring separates durable rules from operational runbooks by migrating command details from MEMORY.md into a dedicated TOOLS.md file. This ensures critical behavioral rules remain preserved while operational instructions become independently accessible.

Normalizing ontology entries structures TODO items and extracts metadata from agent memory files. This process updates the ontology with standardized status tracking for TODOs and enriches document entities to improve searchability and lifecycle management.

Yes, agent memory refactoring integrates with Obsidian by generating extracts from migrated operational content. This creates a searchable, human-readable reference layer for agent tools and runbooks outside the core memory files.

Yes, refactoring agent memory requires Python 3 installed and specific agent directories already present in your environment. These prerequisites enable the scripts to locate, parse, and migrate content between MEMORY.md, TOOLS.md, and the Ontology.

Operational content is automatically moved from MEMORY.md to TOOLS.md when cleaning up documentation for multiple AI agents. The process preserves critical rules in memory while organizing commands and runbooks into dedicated tool files and Obsidian extracts.