# polymnemo MCP server

> Provides shared long-term memory across different LLMs using semantic recall, session management, and multimedia storage via Postgres and pgvector.

- Repository: https://github.com/PCBZ/polymnemo
- Page: https://hostedmcp.org/servers/pcbz/polymnemo
- Transport: remote
- Stars: 0
- License: MIT
- Latest version: 1.5.0

## Install

### Claude Code

```sh
claude mcp add --transport http polymnemo https://polymnemo.purpleground-26f6d744.westus2.azurecontainerapps.io/mcp
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "polymnemo": {
      "url": "https://polymnemo.purpleground-26f6d744.westus2.azurecontainerapps.io/mcp"
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "polymnemo": {
      "type": "http",
      "url": "https://polymnemo.purpleground-26f6d744.westus2.azurecontainerapps.io/mcp"
    }
  }
}
```

### Codex (`~/.codex/config.toml`)

```toml
[mcp_servers.polymnemo]
url = "https://polymnemo.purpleground-26f6d744.westus2.azurecontainerapps.io/mcp"
```

### Gemini CLI (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "polymnemo": {
      "httpUrl": "https://polymnemo.purpleground-26f6d744.westus2.azurecontainerapps.io/mcp"
    }
  }
}
```

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `POLYMNEMO_API_KEYS` | no | no |  |
| `POLYMNEMO_DATABASE_URL` | no | no |  |

## Tools (2)

- `keyword`
- `rrf`

## FAQ

### Does this server make calls to generative LLMs?

No, the server makes no generative-LLM calls and embeddings run locally using ONNX.

### How can I share memories between different AI assistants?

You can point any MCP-capable client at a single polymnemo endpoint to share the same memories stored in your Postgres database.

### Can I store images or videos in the memory?

Yes, you can attach files, images, or video; the actual bytes are sent to object storage while a searchable description is embedded.

### What search modes are available for recalling memories?

The server supports similarity search for meaning, keyword search for exact substrings, and RRF (Reciprocal Rank Fusion) to combine both.