# memory MCP server

> Provides persistent semantic memory for AI agents using PostgreSQL, pgvector, and local Ollama models for hybrid retrieval and memory management.

- Repository: https://github.com/tokarix/memory-server
- Page: https://hostedmcp.org/servers/tokarix/memory-server
- Transport: —
- Stars: 1
- License: Apache-2.0

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `RUST_LOG` | no | no |  |

## Tools (17)

- `memory_server_version`
- `memory_store`
- `memory_search`
- `memory_recall`
- `memory_rules`
- `memory_bootstrap`
- `memory_list`
- `memory_get`
- `memory_update`
- `memory_neighbors`
- `memory_delete`
- `session_start`
- `session_message_append`
- `session_finalize`
- `session_log_store`
- `review_queue`
- `review_submit`

## FAQ

### What technologies power the memory service?

The service is backed by PostgreSQL with pgvector for vector storage and local Ollama models for embeddings and generation.

### How can I share memory across multiple agents?

You can point multiple agents to a central memoryd instance by setting the memoryd_url in their configuration and enabling an api_token for network access.

### Does the server support hybrid search?

Yes, it provides hybrid retrieval using both vector similarity and PostgreSQL full-text search.