# TeamAgentMemory MCP server

> Provides self-hosted, layered long-term memory for AI agents to enable shared context and private conversation isolation within teams.

- Repository: https://github.com/hsuanshao/TeamAgentMemory
- Page: https://hostedmcp.org/servers/hsuanshao/teamagentmemory
- Transport: —
- Stars: 39
- License: Apache-2.0

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `AGENT_MEMORY_LLM_API_KEY` | no | no |  |
| `AGENT_MEMORY_TEST_POSTGRES_URL` | no | no |  |
| `AGENT_MEMORY_TEST_REDIS_URL` | no | no |  |

## FAQ

### What is TeamAgentMemory?

It is a self-hosted long-term memory service for teams working with AI agents that persists context across sessions.

### How does it handle privacy between team members?

It uses a layered approach where individual conversations and facts are private to each member, while distilled scenes and personas are shared by the team.

### Can I run this entirely on my own infrastructure?

Yes, the memory, database, and distillation models can all run within your own network without depending on vendor cloud services.

### What databases are supported?

It supports SQLite for starting out and PostgreSQL with pgvector and Redis for scaling.
