Persistent memory for AI agents — REST API, MCP, OAuth, CLI, dashboard. Semantic search, knowledge graph, automatic consolidation. One self-hosted service, every transport.
Interactive 3D knowledge graph — every memory a glowing node, every relationship a curved edge.
MINOR release: three critical advisories closed. Remote transports served the filesystem tools, the SSE transport had no authentication at all, and open Dynamic Client Registration handed out read-write tokens. Development also moved back to GitHub. ~3,030 tests.
memory_harvest and memory_ingest take a caller-controlled path. The local_only_tools() filter ran in only one of three transports, so Streamable HTTP and SSE served them to remote callers. It now sits in the shared tool list and dispatch, where every transport passes. (GHSA-7crr-2r7w-cpfm)
/sse and /messages/ both reach the full MCP tool surface, and neither was gated. The check existed but was a closure inside another transport's function. Transports now also refuse to start on a non-loopback bind with no authentication configured. (GHSA-2hh8-qjxc-43x3)
With open Dynamic Client Registration, a caller could register itself as confidential, exchange its own credentials, and hold a read-write token the owner never granted. Machine-to-machine grants now require gated registration; authorization_code with PKCE is unaffected. (GHSA-6mvm-q4j3-27qg)
Claude.ai connects to your memory server through Streamable HTTP with OAuth 2.1 authentication.
Battle-tested with comprehensive testing and optimized for performance.
Talks and coverage from the community.
Heavybit's Humans in the Loop deep dive talks to maintainer Heinrich Krupp about agent amnesia and why persistent memory is the missing infrastructure layer for agentic systems.
Maintainer Heinrich Krupp presents mcp-memory-service at the AI Tinkerers Zürich meetup — persistent memory architecture, semantic search, and multi-agent memory sharing. Watch the full video →
Install from PyPI and connect your AI agents to persistent memory.
I help teams run self-hosted memory in production — OAuth, hybrid sync, multi-client rollout. Free 30-minute intro call.
Get in touch