Give your AI agents long-term memory and state persistence using Firestore.
FireMemory is a dedicated MCP server designed to solve the context window limitation of LLMs. It provides a structured way for agents to store important information, recall it when relevant, and maintain a persistent state across different sessions. It runs alongside FireMCP Pro but offers a specialized set of tools focused purely on memory management.
Use this specific URL to connect to the FireMemory MCP server.
FireMemory requires the standard FireMCP headers plus an optional configuration header:
| Header Name | Description |
|---|---|
| x-firebase-api-key | Your Firebase Project API Key |
| x-firebase-project-id | Your Firebase Project ID |
| x-firebase-user-email | Firebase Auth User Email |
| x-firebase-user-password | Firebase Auth User Password |
| x-memory-namespace | Required. Namespace to organize memories (e.g., user_id, project_name) |
| x-memory-collection | Optional. Custom collection name (default: "firemcp_memory") |