Name: Canvas MCP
Author: aryankeluskar

LABSCO SUMMARY

Canvas and Gradescope in one place, which is where the answer usually is.

Coursework state is split across two systems and neither talks to the other; putting both behind one assistant is the actual convenience. Before you install, weigh the credential asymmetry: Canvas gets a revocable access token you generate, while Gradescope takes your account password directly.

What it is

An MCP server that connects an assistant to two things students actually live in: Canvas LMS and Gradescope. It goes past listing due dates — it can find the resource relevant to a question you ask in plain language, and break down what an upcoming assignment involves for a given course.

What you get

Relevant course resources found from a natural-language query, rather than only from a folder you already know to open
Upcoming assignments fetched with a per-course breakdown, not just a flat due-date list
Gradescope courses and assignments queried conversationally, including submission status
The underlying Canvas objects reachable: courses, modules, module items, file URLs, calendar events and assignments

Requirements

Two credentials, both yours to create: a Canvas API key from Canvas > Account > Settings > Approved Integrations > New Access Token, and your Gradescope email and password. Install through Smithery for Claude Code or Cursor with a single npx command, or add it in ChatGPT as a connector pointing at https://canvas-mcp--aryankeluskar.run.tools after enabling Developer Mode. Note that the Gradescope half takes an account password rather than a scoped token, which is a different level of trust from the Canvas key.

Setup effort

One command plus a key — npx -y @smithery/cli@latest mcp add aryankeluskar/canvas-mcp --client claude-code, then supply credentials

See installation options

USE CASES

01

Working out what is due and what it involves

Ask for upcoming assignments and get the per-course breakdown instead of a list of titles and dates.

02

Finding the reading a question refers to

Describe what you are looking for in plain language and let it locate the relevant course resource rather than clicking through modules.

03

Checking whether something submitted

Query your Gradescope courses and assignments for submission status without logging into a second system.

BEFORE YOU INSTALL

What to know first

Requires

npx on PATH

INSTALLATION

Connect your client

Server setup

npx

Claude Code

TERMINAL

npx -y @smithery/cli@latest mcp add aryankeluskar/canvas-mcp --client claude-code

Hosted endpoint

ENDPOINT URL

https://canvas-mcp--aryankeluskar.run.tools

Add this hosted endpoint to your MCP client using the credentials listed under Requirements.

MCP Clients

Cursor

CURSOR · TERMINAL

npx -y @smithery/cli install aryankeluskar/canvas-mcp --client cursor

ChatGPT

Connector url: https://canvas-mcp--aryankeluskar.run.tools

Claude Desktop

CLAUDE DESKTOP · JSON

{
  "mcpServers": {
    "canvas": {
      "command": "npx",
      "args": [
        "-y",
        "@smithery/cli",
        "run",
        "@aryankeluskar/canvas-mcp"
      ]
    }
  }
}

Canvas MCP