Claude Desktop Setup Guide

Quick Setup (Recommended)

Use our automated setup script to configure everything in one step:

curl -sSL https://raw.githubusercontent.com/Kuria-Mbatia/notioc-canvas-mcp-server/main/setup.sh | bash

The script will automatically download, install, configure, and set up the MCP server with Claude Desktop. You'll only need to provide your Canvas URL and API token when prompted.

Prerequisites

Claude Desktop application installed
Node.js 18+ installed on your computer
Canvas LMS access with ability to generate API tokens
Basic familiarity with terminal/command line

Manual Installation

If you prefer to set up manually or the automated script doesn't work for your system:

Step 1: Download the MCP Server

Option A: Clone from GitHub

git clone https://github.com/Kuria-Mbatia/notioc-canvas-mcp-server.git
cd notioc-canvas-mcp-server

Option B: Download ZIP

Download the ZIP file from our GitHub repository and extract it to a directory like /Users/yourusername/notioc-canvas-mcp-server

Step 2: Install Dependencies and Build

Navigate to the MCP server directory and install the required packages:

cd notioc-canvas-mcp-server
npm install
npm run build

Step 3: Get Your Canvas API Token

Log into your Canvas LMS instance
Go to Account → Settings
Scroll down to the "Approved Integrations" section
Click "+ New Access Token"
Enter a purpose like "Kairo MCP Server"
Click "Generate Token"
Copy the generated token (you won't see it again!)

Important: Keep your API token secure and never share it!

Step 4: Configure Environment Variables

Create a configuration file with your Canvas details:

cp .env.example .env

Edit the .env file with your Canvas information:

CANVAS_BASE_URL=https://your-canvas-instance.instructure.com
CANVAS_ACCESS_TOKEN=your-canvas-access-token-here
LOG_LEVEL=info

Step 5: Configure Claude Desktop

Locate your Claude Desktop configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Linux: ~/.config/Claude/claude_desktop_config.json

Add the MCP server configuration to your Claude config file:

{
  "mcpServers": {
    "notioc-canvas": {
      "command": "node",
      "args": ["/absolute/path/to/notioc-canvas-mcp-server/dist/src/server.js"],
      "cwd": "/absolute/path/to/notioc-canvas-mcp-server",
      "env": {
        "CANVAS_BASE_URL": "https://your-canvas-instance.instructure.com",
        "CANVAS_ACCESS_TOKEN": "your-canvas-access-token-here",
        "LOG_LEVEL": "info"
      }
    }
  }
}

Note: Replace /absolute/path/to/notioc-canvas-mcp-server with the actual full path to where you downloaded the MCP server. The cwd parameter sets the working directory for the server. Use forward slashes (/) even on Windows in the JSON configuration.

Step 6: Restart Claude Desktop

Close Claude Desktop completely and restart it. The MCP server should now be available and you'll see

"notioc-canvas" in the list of available tools when you start a new conversation.

Testing Your Setup

Start a new conversation with Claude and try these test commands:

Test Commands:

"What courses am I enrolled in on Canvas?"
"List my current Canvas courses"
"Show me upcoming assignments"
"What assignments do I have for [course name]?"
"Get the syllabus for [course name]"

Success Indicators: You should see Claude mentioning "notioc-canvas" tools in its responses and getting actual data from your Canvas courses.

Troubleshooting

MCP Server Not Recognized

Verify the path in your Claude config is correct and uses absolute paths
Ensure you've run npm run build and the dist/src/server.js file exists
Check that the JSON configuration is valid (use a JSON validator)
Restart Claude Desktop completely (force quit if necessary)
Make sure the cwd parameter points to the correct project directory

Authentication Issues

Verify your Canvas base URL is correct (no trailing slash)
Check that your API token is valid and hasn't expired
Ensure your Canvas instance is accessible from your network
Test your credentials by accessing Canvas API directly in browser: [your-canvas-url]/api/v1/courses

No Data Returned

Check that you're enrolled in active courses
Verify your API token has necessary permissions (course read access)
Look at Claude Desktop logs for error messages
Try different course enrollment states (current, completed, all)

Checking Logs

To see detailed logs for debugging:

macOS: ~/Library/Logs/Claude/
Windows: %APPDATA%\Claude\logs\
Set LOG_LEVEL=debug in your configuration for more verbose output

Path Issues

Common Path Examples:

macOS: /Users/johndoe/notioc-canvas-mcp-server/dist/src/server.js

Windows: C:/Users/johndoe/notioc-canvas-mcp-server/dist/src/server.js

Linux: /home/johndoe/notioc-canvas-mcp-server/dist/src/server.js

Advanced Configuration

Environment Variables

Additional configuration options:

# Required
CANVAS_BASE_URL=https://your-canvas-instance.instructure.com
CANVAS_ACCESS_TOKEN=your-token-here

# Optional
LOG_LEVEL=info              # debug, info, warn, error
REQUEST_TIMEOUT=30000       # Request timeout in milliseconds
CACHE_TTL=300              # Cache time-to-live in seconds
MAX_FILE_SIZE=10485760     # Max file size for reading (10MB)

Complete Configuration Example

Here's a complete Claude Desktop configuration file:

{
  "mcpServers": {
    "notioc-canvas": {
      "command": "node",
      "args": ["/Users/johndoe/notioc-canvas-mcp-server/dist/src/server.js"],
      "cwd": "/Users/johndoe/notioc-canvas-mcp-server",
      "env": {
        "CANVAS_BASE_URL": "https://university.instructure.com",
        "CANVAS_ACCESS_TOKEN": "1234~abcdefghijklmnopqrstuvwxyz123456789",
        "LOG_LEVEL": "info",
        "REQUEST_TIMEOUT": "30000"
      }
    }
  }
}

Available Features

Once configured, you can ask Claude to:

Course Management: List courses, get course details, view enrollments
Assignments: View assignments, check due dates, get submission details
Files & Content: Access course files, read PDFs and documents
Announcements: Get recent announcements and updates
Discussions: View discussion topics and responses
Grades: Check grades and feedback on assignments
Calendar: View upcoming events and deadlines

Usage Examples

Example Conversations:

"What assignments are due this week in my Psychology course?"
"Can you read the syllabus for my CS 101 course and tell me the grading policy?"
"Show me all announcements from my professors this month"
"What's my current grade in Mathematics and what assignments am I missing?"
"Help me organize my upcoming deadlines across all courses"

For more examples and detailed API documentation, visit our MCP Server documentation.

Setup Complete! You should now be able to ask Claude about your Canvas courses and get intelligent

responses based on your actual course data. Claude can help you stay organized, understand course

materials, and manage your academic workload more effectively.

Noti