GH Repository · FlowElement-xinliuyuansu
m_flow
A bio-inspired cognitive memory engine — a new paradigm for Graph RAG.
- stars
- 4,502
- 30-day movement
- +10/day
- Related entries
- 60
- Connections
- 2
mcp-serverpython/uvvector-searchPythonagent-memoryagentic-aiai-reasoningknowledge-graphpythonllmmemory-engineepisodic-memorygraph-databaseragmakedockerlong-term-memorymcpsemantic-memory
m_flow is a Python-based MCP server described as a bio-inspired cognitive memory engine and positioned as a new paradigm for Graph RAG. Based on its topics, it combines knowledge graphs, vector search, and episodic and semantic memory to support LLM long-term memory and reasoning.
Reach for it when you want an MCP-exposed graph-based memory layer for agent or RAG pipelines.
Use it to
- Run as an MCP server for agent memory
- Store and retrieve episodic memory
- Build Graph RAG pipelines with knowledge graphs
- Add long-term semantic memory to LLM apps
- Deploy via the provided Docker/uv toolchain
For Developers building agentic AI systems needing persistent memory
- Role
- mcp-server
- Language
- Python
- Licence
- Apache-2.0
- Forks
- 258
- Last push
- 2026-09-01
- Latest release
- v0.3.4 · 2026-04-12
topicsagent-memorygraph-ragknowledge-graphmcpvector-searchllm