Control Room UI
EdMneme exposes tutoring sessions, project state, tool activity, and infrastructure health through a concrete control-room interface rather than a thin chat wrapper.
Active system
EdMneme
An interactive tutoring stack with a Control Room UI, tool orchestration, project memory, and textbook ingestion. EdMneme is the educational system now represented on the site in place of the older Mneme framing.
Current system
What EdMneme includes today.
The current repository is not a simple tutor prompt wrapped in a UI. It already includes routing, tool orchestration, infrastructure hooks, textbook ingestion, and project-scoped state.
Inference routing and fallbacks
The stack routes across SGLang and Prime/OpenAI-compatible endpoints, keeping tool use and tutoring workflows inside one controllable runtime.
MCP bridge plus guarded local actions
The agent loop validates tool calls through an MCP catalog and dispatches them through local bridges for filesystem, memory, search, vector, verification, and Python execution.
Textbook ingestion and project memory
Remote DeepSeek OCR, chunking, dashboards, and a lightweight knowledge graph give the system a more serious learning substrate than raw chat history.
Why it matters
A tutoring system needs operational depth.
The distinction here is that EdMneme is being built as a real learning system with state, tools, ingestion, and infrastructure visibility. That gives it a more defensible identity than generic “AI tutor” positioning.
Projects, sessions, and prerequisite structure
EdMneme stores projects and sessions with scoped memory so learning can build over time instead of restarting every conversation.
Prime launch and SGLang visibility
The current interface already surfaces Prime offer flows, health checks, streaming logs, and infrastructure state alongside the tutoring surface.
More than chat with a textbook
The product direction is an interactive tutoring system with tool access, structured project state, and deliberate ingestion paths for course material.