MCP (Model Context Protocol)
Connect Anona Memory to Claude Code, Claude Desktop, and other MCP clients so your AI assistant can remember and recall context directly, without you writing any integration code.Tools Exposed
| Tool | What it does |
|---|---|
remember | Store a memory in a space |
recall | Semantically search memories in a space |
list_spaces | List available Memory Spaces |
get_insights | Synthesize an answer from accumulated memories |
Option 1: Local (stdio)
Runs on your machine viauvx, ships with the Python SDK.
Register with Claude Code
ANONA_API_KEY (and optionally ANONA_SPACE_ID) in your shell environment before starting Claude Code.
Option 2: Remote (Streamable HTTP)
Connect directly to the hosted gateway - no local process to run.Important Notes
- Tools only load at session start. Adding or changing the MCP server mid-session won’t expose its tools until you restart your MCP client.
- Scoping to a space. Without
ANONA_SPACE_IDset, tools that need a space will ask you to specify one, or operate across whichever spaces your API key can see. - Same quota, same billing. MCP calls consume credits and count against your usage the same way REST API calls do.
Example: Claude Code Session
Once registered, just talk naturally:“Remember that the user prefers dark mode and uses VS Code.”
“What do we know about this user’s editor preferences?”Claude Code calls
remember and recall automatically behind the scenes.
Next Steps
Python SDK
Programmatic client library
API Reference
Full REST API documentation