Model Context Protocol

Run your academy with your own AI agent

Course37 exposes a secure MCP server, so your organization can connect its own AI agent and do most of what an admin does in the dashboard — add videos, schedule live classes, manage payments, and query exams, finances, students and analytics — all through OAuth and the scopes you choose.

Request a demo See what it can do
Capabilities

What your agent can do

Each capability maps to a tool your agent can call — and each tool is gated by a scope you control.

Courses & content

  • Add videos to any course module
  • Edit course details and pricing
  • Build curriculum / course plans
  • List courses and modules

Live classes

  • Schedule live class sessions
  • Zoom, Google Meet, YouTube Live & streaming
  • Attach sessions to course modules

Payments & finance

  • Record student payments
  • Add installments to a payment
  • Query a student’s dues & history
  • Payment analytics & course revenue breakdown

Profit share

  • Teacher share percentages
  • Expenses & withdrawals
  • Calculate profit share by course & period
  • Per-teacher dashboards

Exams & questions

  • List quizzes/exams & results
  • Merit lists & exam analytics
  • Search the question bank

Students

  • List & search students
  • Student performance analytics
  • Exams, assignments, attendance & video analytics

Website analytics

  • Visitor overview & daily trends
  • Top pages, devices & geography
  • Traffic acquisition & search console

And most of the admin UI

  • The same operations your team does by hand
  • Composable, scope-gated tools
  • New capabilities added over time

How it works

Three steps from zero to an agent that works inside your academy.

Step 1

Generate credentials

In Dashboard → Integrations → MCP Server, create a credential and tick exactly the scopes you want it to have. Write access is off by default.

Step 2

Get a token

Your agent exchanges its client id + secret for a short-lived bearer token using the standard OAuth2 client-credentials grant.

Step 3

Connect your agent

Point any MCP-compatible agent at your MCP endpoint. It can use only the tools your scopes allow — always within your organization.

Exchange credentials for a token
curl -X POST https://api.course37.com/mcp/oauth/token \
  -H "Content-Type: application/json" \
  -d '{
    "grant_type": "client_credentials",
    "client_id": "mcp_live_…",
    "client_secret": "…",
    "scope": "courses:read payments:read analytics:read"
  }'
# → { "access_token": "…", "token_type": "Bearer", "expires_in": 900 }
# then call the MCP endpoint at https://api.course37.com/mcp with that Bearer token
Secure by design

You stay in control

The agent only ever has the access you grant — nothing more. Tighten or revoke it the moment you need to.

  • OAuth2 client-credentials — no passwords shared with the agent
  • Every tool is gated by an explicit scope you grant
  • Write & destructive tools are off until you enable them
  • Strictly organization-scoped — an agent only ever sees your data
  • Secrets are hashed at rest and shown only once
  • Disable or rotate a credential at any time — effective immediately

Give your AI agent the keys — safely

Set it up in minutes from your dashboard under Integrations → MCP Server.

Request a demo Open Integrations