Axial Coach
MCP server that exposes a student's cognitive profile as read-only tools for Sauna, enabling proactive coaching with gamified quests and holistic parent reports.
README
Axial Coach — a cognitive coworker on Sauna
Sauna × ElevenLabs hackathon. A coworker that doesn't just remember what a student did — it understands how they think, and coaches on it: proactively, for the whole support system around the learner, in voice.
Axial maps how a student thinks — a set of connected cognitive skills across six domains, scored from a diagnostic. The scoring is deterministic and explainable: every insight traces back through typed skill-to-skill dependencies (the "because-trail"). No AI decides a score.
This project connects Axial to Sauna as a custom MCP connector, so Sauna can reason and act on a student's real cognitive profile from any surface — and gives it a voice with ElevenLabs.
Axial gives the substance. Sauna gives the hands. ElevenLabs gives the voice.
What it does
Once Axial is connected to Sauna (one URL — see below), Sauna can:
- Explain, grounded. "What should this student work on first, and why?" → Sauna answers from the real leverage + because-trail — naming the specific skills a focus unlocks. Nothing hallucinated.
- Serve two audiences from one model. The same profile becomes:
- 🎮 A gamified quest for the student — a weekly mission, small daily wins with XP and a streak, encouragement anchored to a real strength. (Confidence through constant small wins.)
- 🌱 A holistic report for the parent — status across all six thinking domains plus the week's focus and its cross-domain ripple — evidence that one small habit grows skills in multiple areas. (Holistic development, not one score.)
- Run proactively. A Sauna scheduled task re-pulls the live Axial profile every week and regenerates both — working while you're offline, not waiting to be asked.
- Speak. In Sauna, the coach answers in voice (ElevenLabs).
How it's built
┌───────────────────────────────┐ Sauna (web / iMessage / Slack / email)
│ Axial engine (proprietary, │ • connect via one MCP URL
│ NOT in this repo) │ • calls the tools, reasons, acts
└──────────────┬────────────────┘ • speaks the answer (ElevenLabs)
│ run once, offline ▲
▼ │ MCP (JSON-RPC over HTTP)
maya-axial.json (precomputed, ┌────────────┴─────────────┐
citation-free snapshot, 1 student) ──▶│ server.mjs POST /mcp │
│ 6 read-only tools │
└──────────────────────────┘
The Axial engine runs once, offline, and its output for one sample student is
cached in maya-axial.json. server.mjs serves slices of that snapshot over MCP —
so the coworker gets Axial's real, explainable intelligence without the engine or
taxonomy ever leaving Axial.
MCP tools (all read-only, plain-language, no internal ids):
get_cognitive_profile, get_bottlenecks (with because-trail), recommend_methods,
get_weekly_focus, get_student_quest (gamified), get_parent_report (holistic 6 domains).
Run it
node server.mjs # starts http://localhost:8787 (/mcp)
cloudflared tunnel --url http://localhost:8787 # get a public https URL
Then in Sauna: Connections → Add a Connection → MCP Server → paste https://<tunnel>/mcp.
Sauna probes the server, discovers the tools, and you can ask it about the student.
(The optional POST /tts route is a standalone ElevenLabs proxy; it needs
ELEVENLABS_API_KEY in the environment. In the demo, voice comes from Sauna's
built-in ElevenLabs, so this route isn't required.)
Sponsors
- Sauna — the multiplayer AI coworker. Custom MCP connector; proactive scheduled runs; the two-audience coaching; voice.
- ElevenLabs — the coach's voice.
A note on IP
Axial's scoring engine, its full cognitive taxonomy, and the cited research behind it are proprietary and deliberately not included here. This repository contains only the Sauna integration and a precomputed, citation-free snapshot for one sample demo student (no real student data, no effect sizes, no sources). It is a hackathon demo, not the Axial product.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.