Integrate Decentralized Memory for AI Agents

Membase MCP Server connects an AI agent to Membase, described as the first decentralized memory layer for AI agents, powered by Unibase. It uploads and retrieves conversation history, interaction records, and knowledge to the Unibase DA network, giving agents persistent, verifiable, and traceable memory across sessions instead of storage that disappears when a session ends.

Use it when you want an agent's conversation memory to persist on a decentralized, verifiable storage layer - useful for continuity across sessions, personalization, or auditability of what an agent said and stored. Each conversation is tracked by a unique MEMBASE_CONVERSATION_ID, and stored messages/memories can be inspected directly at testnet.hub.membase.io. Do not use it if you need purely local, private memory with no external network dependency - Membase stores data on the Unibase DA network rather than locally.

• get_conversation_id: get the ID of the current conversation.
• switch_conversation: switch to a different conversation, preloading its history.
• save_message: save a message or memory into the current conversation.
• get_messages: retrieve the last n messages from the current conversation.

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

Configure required environment variables: MEMBASE_ACCOUNT (your account to upload as, e.g. 0x...), MEMBASE_CONVERSATION_ID (a unique conversation identifier whose history will preload), and MEMBASE_ID (your instance/sub-account identifier).

Register with Claude, Windsurf, Cursor, or Cline:

{
  "mcpServers": {
    "membase": {
      "command": "uv",
      "args": ["--directory", "path/to/membase-mcp", "run", "src/membase_mcp/server.py"],
      "env": {
        "MEMBASE_ACCOUNT": "your account, 0x...",
        "MEMBASE_CONVERSATION_ID": "your conversation id, should be unique",
        "MEMBASE_ID": "your sub account, any string"
      }
    }
  }
}

Agent developers who want persistent, decentralized, and independently verifiable conversation memory for their AI agents rather than storage tied to a single centralized provider.