purmemo

The pūrmemo MCP server provides a persistent identity and memory layer for your MCP-enabled AI sessions. It stores your identity, tools, active projects, and conversation history so that each new session can resume with context relevant to you. The server exposes memory and context management capabilities that downstream MCP clients can query, including loading your identity profile, accessing recent conversations, and saving new memories or conversations. Tools available through the server include get_user_context (to load your identity and current context), save_conversation (to store ongoing dialogues with meaningful titles), recall_memories (to search memories by topic), and related-conversations discovery to find connected discussions across platforms. This enables cross-session continuity and richer interactions across Claude, Cursor, Windsurf, Zed, and other MCP-compatible clients.

To use pūrmemo with a client, configure the MCP server in your environment (e.g., Claude Desktop, Cursor, Windsurf, or Zed) to point at your PURMEMO MCP server entry, typically via the mcpServers object with the server name purmemo. Once the server is reachable, the client can request your memory and identity data and perform memory operations as part of session startup or ongoing conversations.

Prerequisites:

• Node.js v14+ and npm installed on your system
• Access to install npm packages (registry access)
• Optional: PURMEMO_API_KEY if you are using Purmemo's API-backed features

Installation steps:

1. Prerequisites check

• Ensure Node.js and npm are installed:

node -v
npm -v

2. Run the MCP server via npx (quick start)

• The recommended quick-start command uses npx to fetch and run the purmemo-mcp server:

npx -y purmemo-mcp

• If you need to provide an API key, set the environment variable PURMEMO_API_KEY before running:

export PURMEMO_API_KEY=your-api-key-here
npx -y purmemo-mcp

3. Local installation (optional)

• Install the package locally in a project to pin versions:

npm init -y
npm install purmemo-mcp --save

• Start the server locally (adjust path as needed depending on your setup):

node node_modules/purmemo-mcp/dist/index.js

4. Docker (alternative, if supported in your environment)

• If you prefer Docker, you can run the image (replace image-name with the actual image if published):

docker run -i purmemo-mcp

5. Verify and connect

• Once running, configure your MCP client to point at the purmemo MCP server endpoint (e.g., the URL provided by your hosting setup) and ensure the API key, if required, is set in environment variables.

Tips and common issues:

• If using the npx approach, ensure you have network access to fetch the package from npm during first run.
• When using an API key, keep it secure and do not hard-code it into public configuration files.
• For Claude Desktop or IDE integrations, update the mcpServers configuration to include the purmemo server with the appropriate command and args shown in the README examples.
• If you rotate API keys, remember to update the PURMEMO_API_KEY environment variable across all clients using the MCP server.
• The MCP server works best when your memory and identity data are kept fresh; periodically review and prune outdated memories to maintain performance.
• If you encounter connectivity issues, verify that the MCP server URL is reachable from the client and that any required OAuth or authentication flows are completed as described in the client’s setup steps.