recall

Recall is a persistent brain extension that stores memories across sessions for Claude and other AI agents. When you run Recall in self-hosted mode (with Redis or Valkey), memories are saved locally and can be retrieved automatically in future conversations, even after restarts or context window resets. You can organize memories by workspaces, enabling project-specific memories that don’t spill over into other projects. To use Recall, run the MCP server on your infrastructure, connect your Claude agent or other clients to the MCP endpoint, and begin storing memories via simple prompts like “Store a memory that I prefer TypeScript for all projects.” You can also retrieve context-aware memories automatically as conversations progress, thanks to the auto-consolidation and cross-session memory capabilities. If you opt for the cloud version, you connect Claude Desktop or other clients to recallmcp.com with your API key to enable seamless memory syncing without managing infrastructure.

Prerequisites:

• Node.js v18 or later installed on your server
• Redis 6.x+ or a Valkey instance accessible from the server
• Git (optional, for cloning the repo)

Install steps:

1. Clone the repository (if you have a local copy): git clone https://github.com/your-org/recall-mcp.git cd recall-mcp

2. Install dependencies (if applicable): npm install

3. Configure environment:

   • Ensure Redis is accessible (REDIS_URL)
   • Set VALKEY_API_KEY if using Valkey
   • Provide any required embedding provider keys if necessary

4. Run the MCP server: npm run build node dist/server.js

5. Verify the server is up by hitting the MCP endpoint from a client: http://<your-server-host>:<port>/mcp

Notes:

• In production, use a process manager like PM2 or systemd to keep the server running.
• If using Redis/Valkey authentication, ensure TLS and proper firewall rules are in place.

Environment tips:

• Use REDIS_URL with TLS (rediss://) in production for Redis.
• Prefer separate Redis databases or Valkey namespaces per environment (dev/stage/prod).
• Ensure API clients (Claude Desktop or other) are configured to point to your MCP endpoint and have proper authentication if required.
• Enable memory pruning or retention policies according to your data governance requirements.
• Regularly monitor memory usage and storage to avoid unbounded growth.
• If you switch from cloud to self-hosted, migrate memories by exporting/importing via the MCP interface if supported.