memory-skills

Structured memory management for AI agents organizes knowledge into working, episodic, semantic, and procedural memories to enable recall and context resume in long-running conversations. It uses the CoALA framework to separate project-level and global memory for scalable knowledge continuity.

The CoALA framework defines a memory architecture for language agents that categorizes knowledge into working, episodic, semantic, and procedural memories. It enables organized long-term and short-term memory management with clear data flows for development workflows.

To save and recall session context, use commands like /save-context to store the current session and /recall to retrieve relevant memories. You can also use /reflect to convert episodic memories into semantic insights for future use.

No external dependencies are required to implement episodic and semantic memory. The system provides a structured memory taxonomy with persistent storage paths and safe memory API interaction natively without additional packages.

Yes, you can separate project-level and global memory in AI agent workflows. This separation allows scalable knowledge management by isolating specific project contexts from broader global knowledge across different development sessions.