Canvas LMS MCP Server

A Model Context Protocol (MCP) server that provides Claude with secure, verified access to your Canvas LMS account at Texas Tech University.

What is MCP?

The Model Context Protocol (MCP) is an open protocol developed by Anthropic that allows AI assistants like Claude to securely interact with external services. Think of it as "USB-C for AI" — a standardized way to connect Claude to your data and tools.

This MCP server enables Claude to:

• 📚 List your enrolled courses

• 📝 Retrieve assignments and due dates

• 📊 Check your grades

• 📢 Read course announcements

• 🗓️ View upcoming events and to-do items

• 💬 Access discussion topics

Features

• Test-First Design: Only exposes endpoints verified to work with your Canvas account

• Secure: API tokens stored in .env, never committed to version control

• Type-Safe: Full Pydantic validation on all inputs

• Actionable Errors: Clear error messages guide you to solutions

• Dual Output: Supports both human-readable Markdown and machine-readable JSON

Prerequisites

• Python 3.10+

• uv (recommended) or pip

• Canvas API Token from your institution

• Claude Desktop or Claude Code (for MCP integration)

Installation

1. Clone the Repository

2. Install Dependencies

Using uv (recommended):

Using pip:

3. Configure Credentials

Copy the example environment file:

Edit .env with your Canvas credentials:

How to Get Your Canvas API Token

1. Log in to Canvas at https://texastech.instructure.com

2. Click your profile picture → Settings

3. Scroll to Approved Integrations

4. Click + New Access Token

5. Enter a purpose (e.g., "Claude MCP Server")

6. Click Generate Token

7. Copy the token immediately — you won't see it again!

4. Configure Test Hints (Optional)

For targeted testing, create test_hints.json:

Running the Server

Option A: Direct Execution (Testing)

Option B: With MCP Inspector (Debugging)

The MCP Inspector provides a web UI to test your server:

Then open http://localhost:8000/mcp in the Inspector.

Option C: Claude Desktop Integration

Add to your Claude Desktop configuration (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

Important: Use absolute paths. On Windows, use forward slashes or escaped backslashes.

Restart Claude Desktop completely (Cmd+Q on macOS, not just close the window).

Option D: Kiro CLI Integration

For integration with Kiro CLI, add the server to your MCP configuration:

1. Install the Canvas LMS MCP Server:

   git clone https://github.com/sweeden-ttu/canvas-lms-mcp.git cd canvas-lms-mcp python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate pip install -e .

2. Configure your Canvas API token:

   cp .env.example .env # Edit .env with your Canvas API token and base URL

3. Add to Kiro CLI MCP configuration (~/.kiro/mcp_servers.json):

   { "mcpServers": { "canvas-lms": { "command": "/ABSOLUTE/PATH/TO/canvas-lms-mcp/venv/bin/python", "args": ["/ABSOLUTE/PATH/TO/canvas-lms-mcp/server.py"], "env": {} } } }

4. Restart Kiro CLI to load the new MCP server.

Important: Use absolute paths and ensure the .env file contains your Canvas API token.

Option E: Claude Code Integration

Available Tools

Once connected, Claude can use these tools:

Example Usage with Claude

Once configured, you can ask Claude things like:

"What assignments are due this week in my Canvas courses?"

"Show me the announcements from all my classes"

"What's my current grade in course 58606?"

"List all my active courses"

"What do I have on my to-do list?"

Querying Endpoints Manually

For debugging or scripting, you can query the Canvas API directly:

Project Structure

Troubleshooting

Server Not Appearing in Claude Desktop

1. Check JSON syntax: Validate claude_desktop_config.json in a JSON linter

2. Use absolute paths: Relative paths won't work

3. Restart completely: Cmd+Q (macOS) or right-click system tray → Quit (Windows)

4. Check logs: ~/Library/Logs/Claude/mcp*.log (macOS)

401 Unauthorized Errors

• Your API token may be invalid or expired

• Regenerate a new token in Canvas Settings → Approved Integrations

403 Forbidden Errors

• Some endpoints require instructor/TA privileges

• The /files endpoint typically requires elevated permissions

• This is normal for student accounts

Rate Limiting (429)

• Canvas has rate limits (typically 700 requests per 10 minutes)

• The server implements exponential backoff automatically

• If you hit limits, wait a few minutes before retrying

Connection Refused

• Ensure the server is running

• Check that the path in the config is correct

• Verify Python and uv are in your PATH

Development

Running Tests

Generating the Specification

After running live tests:

This creates verified_canvas_spec.json documenting which endpoints work.

Code Quality

Security Considerations

• Never commit — it contains your API token

• Token scope: Canvas tokens have full access to your account; treat them like passwords

• Local only: This server runs locally via stdio; it doesn't expose an HTTP endpoint by default

• No persistence: The server doesn't store any Canvas data

Contributing

1. Fork the repository

2. Create a feature branch (git checkout -b feature/amazing-feature)

3. Run tests (uv run pytest)

4. Commit changes (git commit -m 'Add amazing feature')

5. Push to branch (git push origin feature/amazing-feature)

6. Open a Pull Request

License

This project is licensed under the MIT License — see the LICENSE file for details.

Acknowledgments

• Anthropic for creating the Model Context Protocol

• Canvas LMS for their comprehensive REST API

• FastMCP for the Python MCP SDK

• Amazon Web Services for Kiro CLI integration support

Integration Documentation

• Kiro CLI Integration: See installation instructions above for Kiro CLI setup

• Claude Desktop Integration: See CLAUDE.md for detailed Claude Desktop setup

• Amazon Q CLI Integration: See QCHAT_INTEGRATION.md for Q CLI setup

Links

• MCP Documentation

• Canvas REST API Documentation

• Claude Desktop

• Claude Code