Research multi-agent + MCP server

RLM Agent

Multi-agent scientific-literature research with persistent memory, exposed to any coding agent as an MCP server. Public, working code.

RLM Agent technical proof

RLM Agent public research fixture running in Terminal with cited findings and stream metadata.
The problem
Research conclusions are hard to reuse when papers, code, agent traces, and prior findings live in separate places.
Who it is for
Developers and researchers who need literature research inside an MCP-compatible coding tool.
What the system proves
A public local MCP server with persistent research memory, transparent degradation, and 685 tests.

Technical proof

Open the request path.

Inspect each stage, the code evidence behind it, and the limits the system keeps visible.

Request path — verified against source
click any stage · arrow keys work too