Canvas AI Assistant with MCP

Overview

This repository contains our project for building a Model Context Protocol (MCP) server that integrates with the Canvas learning management system (LMS). The goal is to provide an AI-driven assistant (e.g., Claude) with structured access to course information—assignments, due dates, lecture notes—so that it can help students stay organized and succeed academically.

Getting Started

Mac:

0. Install uv if you haven't already:

# if you're paranoid about piping to sh, go read the script by copy pasting the URL
curl -LsSf https://astral.sh/uv/install.sh | sh

1. Clone the repository:

git clone https://github.com/AdityaPrakash-26/Canvas-MCP.git && cd Canvas-MCP

2. Create the virtual environment:

uv venv --seed

3. Activate the virtual environment:

source .venv/bin/activate

4. Install the dependencies:

uv sync

Installing to Claude Desktop

If you don't have Claude Desktop MCP setup, you can follow the Quickstart Guide to install it.

To install this MCP server in Claude Desktop:

1. Open Claude Desktop
2. Go to Settings > Advanced > Edit Configuration
3. (MAC ONLY) Add the following to your claude_desktop_config.json file in the tools section: (it is a key under mcpServers)
4. (WINDOWS ONLY) See instructions below

REPLACE $DIR with the absolute path to the directory where you cloned this repo. (MANDATORY!!!!!) REPLACE $DIR_uv with the uv path. (MANDATORY!!!!!) you can find it by running

which uv

"Canvas MCP": {
      "command": "$DIR_uv",
      "args": [
        "run",
        "--with",
        "canvasapi>=3.3.0",
        "--with",
        "mcp[cli]",
        "--with",
        "python-dotenv>=1.0.1",
        "--with",
        "structlog>=24.1.0",
        "--directory",
        "$DIR",
        "src/canvas_mcp/server.py"
      ]
    }

Windows

We recommend using WSL (Windows Subsystem for Linux) to use this software. Installation instructions can be founder here.

• Once installed, set up a new user in /home/<username>
• Once you have the virtual environment setup, modify your claude desktop app config to include the following:

"Canvas MCP": {
  "command": "wsl.exe",
  "args": [
    "-d",
    "Ubuntu",
    "--exec",
    "/home/<USER>/.local/bin/uv",
    "run",
    "--with",
    "canvasapi>=3.3.0",
    "--with",
    "mcp[cli]",
    "--with",
    "python-dotenv>=1.0.1",
    "--with",
    "structlog>=24.1.0",
    "--directory",
    "/home/<USER>/Canvas-MCP",
    "src/canvas_mcp/server.py"
  ]
}

Replace <USER> with your username.

Running

To run the mcp command with uv:

uv run mcp

Testing

Alternatively, you can test it with the MCP Inspector:

mcp dev src/canvas_mcp/server.py

Secrets

Clone .env.example to .env and add your Canvas API key. Also change the variables in the .env file to match your Canvas instance. "CANVAS_API_KEY" "CANVAS_API_URL"

---

Repository & Kanban

• This public GitHub repository contains our code, documentation, and issues/tasks.
• We maintain a Kanban board under the “Projects” tab with tasks from our roadmap for the first sprint.

Link to the Repo: (https://github.com/AdityaPrakash-26/Canvas-MCP)

Link to the Kanban: (https://github.com/users/AdityaPrakash-26/projects/1)

---

License