# unibaseio/membase-mcp [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/unibaseio/membase-mcp  
**GitHub Stars:** 24  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/unibaseio-membase-mcp

## Description
Save and query your agent memory in distributed way by Membase

## Claude Desktop Quick Installation
Heuristic fallback — verify the package name and runner against the repository README before running it. Uses `npx` (confidence: low):

```json
"mcpServers": {
  "membase-mcp": {
    "command": "npx",
    "args": ["-y","unibaseio-membase-mcp"],
    "env": {
      "MEMBASE_ACCOUNT": "",
      "MEMBASE_CONVERSATION_ID": "",
      "MEMBASE_ID": ""
    }
  }
}
```

**Requires environment variables:** `MEMBASE_ACCOUNT`, `MEMBASE_CONVERSATION_ID`, `MEMBASE_ID` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What unibaseio/membase-mcp MCP server does

The unibaseio/membase-mcp MCP server gives an MCP-compatible agent access to Membase memory. It is intended for storing conversation history, interaction records, and other agent memories in the Unibase DA network. The stored data can be visited through the Membase testnet hub at https://testnet.hub.membase.io/.

Memory is organized around conversations. The server keeps an active conversation context, lets the agent identify that context, and supports switching to another conversation. Messages can then be saved to the selected conversation or retrieved from it.

## How it works

The server is launched as a local Python entry point through `uv`. At startup, configuration identifies the account used for uploads, the conversation whose history should be preloaded, and the instance identity. The documented variables are `MEMBASE_ACCOUNT`, `MEMBASE_CONVERSATION_ID`, and `MEMBASE_ID`.

The unibaseio/membase-mcp MCP server exposes these operations through MCP tool calls rather than requiring the agent to implement direct Membase protocol requests. A client can first obtain the current conversation ID, switch conversations when needed, save a message or memory, and request the last number of messages from the current conversation.

## Setup and configuration

Clone the repository, enter its directory, and start the server with:

```shell
git clone https://github.com/unibaseio/membase-mcp.git
cd membase-mcp
uv run src/membase_mcp/server.py
```

Set `MEMBASE_ACCOUNT` to the account used for uploading data. Set `MEMBASE_CONVERSATION_ID` to a unique conversation ID; the README states that this conversation's history is preloaded. Set `MEMBASE_ID` to the instance or sub-account identifier. The client configuration uses the `uv` command with the repository directory, followed by `run src/membase_mcp/server.py`.

The README lists Claude Desktop, Windsurf, Cursor, and Cline configuration examples. The directory path and environment values in that configuration must be replaced with values for the local installation.

## Tools and capabilities

The unibaseio/membase-mcp MCP server provides four documented functions:

- `get_conversation_id` returns the current conversation ID.
- `switch_conversation` changes the active conversation.
- `save_message` stores a message or memory in the active conversation.
- `get_messages` retrieves the last requested number of messages from the active conversation.

Together, these functions support loading an existing conversation, changing the memory context, recording new information, and reading recent history.

## Limitations and notes

The provided material documents the server against the Membase testnet hub and does not describe production deployment, retention controls, access policies, message schemas, or synchronization behavior. It also does not specify an API-key authentication flow or provide a package-manager distribution command. Configuration therefore depends on running the repository's source entry point and supplying the three documented environment variables.

_Full upstream README: https://allmcps.com/mcp/unibaseio-membase-mcp/readme_