[FEATURE] Add MemMachine memory tool integration #396
Description
Problem Statement
Add a memmachine_memory tool that integrates MemMachine as a memory backend for Strands agents.
MemMachine is an open-source long-term memory layer for AI agents that provides episodic (conversational) and semantic (factual) memory types. MemMachine supports both a hosted cloud platform (api.memmachine.ai) and self-hosted deployments.
Proposed Solution
A new tool memmachine_memory with 4 actions mapped to the MemMachine v2 API:
| Action | API Endpoint | Description |
|---|---|---|
| store | POST /v2/memories | Store memories with metadata, producer context, and type selection |
| search | POST /v2/memories/search | Semantic search with top-k, type filtering, and metadata filtering |
| list | POST /v2/memories/list | Paginated listing with type and metadata filters |
| delete | POST /v2/memories/episodic/delete or /semantic/delete | Single or bulk delete by memory ID |
Use Case
Agents can store and recall information across sessions — like user preferences, past conversations, and accumulated knowledge — so users don't have to repeat themselves. For example, a travel agent remembers "I prefer aisle seats" from a previous chat and applies it automatically when booking a new flight. Works with both the hosted MemMachine platform and self-hosted deployments.
Alternatives Solutions
No response
Additional Context
No response