Connect Claude to 20 years of teaching innovation.
Every method answer Larry gives comes from the Brain: a live map of 181 thinking methods, which one fits this moment, and how they chain, calibrated by what actually worked for 100+ real ventures. Your key keeps that connection alive, and answers say where they came from.
Free 30-day trial. No credit card. Key delivered instantly.
From sign-in to Brain-connected.
Sign in
One-click Google. Your 30-day trial starts instantly.
Get the key
Unique API key provisioned automatically. Copy from your dashboard.
Connect
One config entry. Brain intelligence activates immediately.
Grounded answers, or an honest refusal.
Larry's method answers come from the teaching graph and say so, with a source line. Chaining, calibrated grading against 100+ real projects, and cross-domain patterns all ride on the same connection.
Larry tells you plainly the Brain is disconnected and refuses to invent method answers, instead of serving a look-alike. Your rooms, files, and local graph keep working the whole time. A refusal is the product being honest, not broken.
Plans and prices live on one page: see pricing. The trial is free, no credit card.
Got your key? Connect in 60 seconds.
Two paths, depending on where you run Claude.
Run this
/mos:setup brain # Larry asks for your API key # Paste it when prompted # Done. Brain is connected.
Or manually: add MINDRIAN_BRAIN_KEY=your-key to your project's .env.
claude_desktop_config.json
{
"mcpServers": {
"mindrian-brain": {
"url": "https://mindrian-brain.onrender.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Replace YOUR_API_KEY with the key from your dashboard. Restart Claude Desktop.
If something feels off.
Getting 'RESOURCE_EXHAUSTED' or a Pinecone quota error?
Run rm -f .mcp.json then restart. Your rooms, files, and local graph are untouched; while the Brain is unreachable, Larry says so and holds off on method answers instead of improvising.
Getting '401 Invalid API key'?
Check the key. If expired, sign in again to verify trial status.
Brain server slow on first request (30-60s)?
The Brain server sleeps after 15 minutes idle. The first call wakes it. Subsequent calls are fast.
Old config with neo4j-brain / pinecone-brain entries?
Delete .mcp.json and use /mos:setup brain instead. The old direct-database pattern is deprecated.
Get your key.
Sign in with Google. 30-day free trial. Key delivered instantly.