# omem MCP server

> Provides AI agents with shared persistent memory across sessions, devices, and teams using a space-based architecture.

- Repository: https://github.com/ourmem/omem
- Page: https://hostedmcp.org/servers/ourmem/omem
- Transport: —
- Stars: 174

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `AWS_ENDPOINT_URL` | no | no |  |
| `AWS_REGION` | no | no |  |
| `OMEM_API_KEY` | no | no |  |
| `OMEM_API_URL` | no | no |  |
| `OMEM_OSS_BUCKET` | no | no |  |
| `OMEM_S3_BUCKET` | no | no |  |
| `OSS_ACCESS_KEY_ID` | no | no |  |
| `OSS_ACCESS_KEY_SECRET` | no | no |  |
| `OSS_ENDPOINT` | no | no |  |

## Tools (15)

- `memory_store`
- `memory_search`
- `memory_get`
- `memory_update`
- `memory_delete`
- `memory_list`
- `memory_ingest`
- `memory_profile`
- `memory_stats`
- `space_create`
- `space_list`
- `space_add_member`
- `memory_share`
- `memory_pull`
- `memory_reshare`

## FAQ

### How does omem handle memory lifecycle?

It uses a Weibull decay model to manage the memory lifecycle, where core memories persist and peripheral ones gracefully fade without manual cleanup.

### What kind of search capabilities does it offer?

It features an 11-stage hybrid retrieval pipeline including vector search, BM25, RRF fusion, cross-encoder reranking, and MMR diversity.
