jido-memory

To define memory scope for AI agents, you specify what data to remember—such as conversations, tool results, and domain facts—and establish explicit rules for when to write or read memory across Jido workflows. This ensures agents retain relevant long-term context.

The best way to separate memory logic from core agent behavior is to enforce strict boundaries that keep memory semantics within the memory module. This approach provides explicit interfaces for storage backends while maintaining clear isolation of agent logic.

You set up long-term memory storage by defining explicit persistence interfaces for external storage backends. This separates storage policy from implementation, allowing Jido workflows to write and read episodic traces and tool results during execution.

An agent should write or read memory based on predefined memory scope decisions that specify timing within the workflow. This determines when to persist episodic traces or tool results and when to retrieve domain facts to inform agent behavior.