An MCP server implementing Recursive Language Models (RLM) to process arbitrarily large contexts through a programmatic probe, recurse, and synthesize loop. It enables LLMs to perform multi-step investigations and evidence-backed extraction across massive file sets without being limited by standard context windows.