Definition
Declarative memory in the context of AI agents refers to the explicit storage of facts, rules, and structures that an agent can query and reason over. This is distinct from the implicit weights of the model or the transient history of a single session.
Why it matters (in Poovi’s context)
A critical component for avoiding “agent amnesia.” Tools like graphify (via knowledge graphs) and agentic_stack (via the .agent folder) implement declarative memory to provide persistence across sessions and tools.
Key properties or components
- Persistence: Lives beyond a single chat session.
- Queryability: Can be searched or retrieved via specific patterns.
- Portability: Can be moved between different agent interfaces.