MCP Server Documentation

What is Teach Me Something?

Teach Me Something is a personal AI tutor that connects to Claude, Claude Code, or ChatGPT via the Model Context Protocol (MCP). It gives your AI a persistent library of tutorials tailored to your background, a learner profile that tracks your goals and experience, and a spaced repetition quiz system so you actually retain what you learn.

Everything is stored on the server. Switch between AI clients and your learning history follows you.

Features

• Learner profile — your AI interviews you about your background, goals, and learning style, then uses that context for every tutorial it writes.
• Subject organization — group tutorials under subjects like "Rust", "Biology", or "Music Theory", each with its own background context.
• Personalized tutorials — your AI writes tutorials that reference your actual projects and build on what you already know. Tutorials support markdown, Q&A sections, concept tags, and prerequisites.
• Margin notes — jot down notes while reading tutorials in the web UI, then ask your AI to review and incorporate them.
• Spaced repetition quizzes — a Fibonacci-based algorithm schedules reviews right before you'd forget. Understanding scores (1-10) update automatically after each quiz.
• Publish and share — make any tutorial public with a shareable URL.

Setup instructions

First, create an account on Teach Me Something. Then connect your AI using one of the methods below.

Claude.ai / Claude Desktop

1. In Claude, go to Customize → Connectors → Add Custom Connector.
2. Enter the name Teach Me Something and the URL https://teachmesomething.xyz/mcp.
3. Enter the Client ID and Client Secret shown on your Settings page under the "Claude Web" tab.
4. Click Connect and follow the on-screen OAuth flow to authorize.

No API token needed — OAuth handles authentication automatically.

Claude Code

1. Go to your Settings page and create an API token under the "API Tokens" section.
2. Copy the token (it is only shown once).
3. Run this command in your terminal:

   claude mcp add --transport http \
     teach-me-something https://teachmesomething.xyz/mcp \
     --header "Authorization: Bearer YOUR_API_TOKEN" \
     --scope user

4. Restart Claude Code, then run /mcp to verify the connection.

ChatGPT

1. In ChatGPT, go to Apps → Advanced Settings → enable Developer Mode → Create App.
2. Enter the name Teach Me Something and the MCP Server URL https://teachmesomething.xyz/mcp.
3. Set authentication to OAuth and enter the Client ID and Client Secret shown on your Settings page under the "ChatGPT" tab.
4. Check the acknowledgment checkbox and click Create.

Authentication

The MCP server at https://teachmesomething.xyz/mcp requires a Bearer token on every request. There are two ways to authenticate:

• OAuth (Claude.ai, ChatGPT) — the AI client handles the OAuth flow automatically. You authorize via a browser redirect and never handle tokens directly.
• API tokens (Claude Code) — generate a token on the Settings page. The token is shown once at creation. Pass it as a Bearer token in the Authorization header. Tokens are stored as SHA-256 hashes and can be revoked at any time from Settings.

Usage examples

Example 1: Start learning a new subject

You type into Claude or ChatGPT:

Teach me something about Rails. I've heard that it's very beginner friendly.

Your AI creates a "Rails" subject, interviews you about your background — what programming languages you know, whether you've done web development before, what you want to build. Your answers get saved to the subject's background so every future Rails tutorial is personalized to you. Then it writes your first tutorial based on what you told it.

Example 2: Learn from your own project

You're building a blog app in Claude Code and type:

I added a posts controller but I don't understand how the URL gets to my code. How does that work?

Your AI looks at where you are in your project, scans the relevant files, and creates a tutorial about routing that uses your actual blog app as the running example. It saves it to your Rails subject and it shows up in your web library immediately.

Example 3: Quiz yourself on weak topics

You type:

Quiz me on whatever I need to review most.

Your AI checks the spaced repetition schedule, picks the tutorials you're most likely to forget, and quizzes you. After the quiz, your understanding score updates automatically. Topics you're shaky on come back in a couple of days; topics you know well won't come back for weeks.

Example 4: Review and incorporate margin notes

You're reading your routing tutorial in the web UI and highlight a section with the note "wait, what's the difference between resources and resource?" Then you type:

I left some notes on my routing tutorial. Can you address them?

Your AI reads your margin notes, updates the tutorial to address your questions, and clears the notes once they've been incorporated.

Support

For questions, bug reports, or feature requests, email piyushagarwal5050@gmail.com.