We're excited to announce the open-source release of Canvas MCP, an MCP server that bridges AI coding agents and Canvas LMS. Available on PyPI and the MCP Registry, Canvas MCP enables Claude, Cursor, Codex, and 40+ other AI agents to manage courses through natural language.
Why Canvas MCP?
Instructors spend significant time on repetitive LMS tasks: creating assignments, grading submissions, managing modules, and sending announcements. Canvas MCP automates these workflows by letting AI agents interact with Canvas directly, turning natural language instructions into API calls.
What's Included
- 90+ tools covering assignments, grades, discussions, modules, pages, rubrics, peer reviews, and more
- 8 agent skills for complex multi-step workflows like course audits and bulk grading
- 290+ tests ensuring reliability across Canvas API endpoints
- Multiple distribution channels — PyPI, MCP Registry, skills.sh, and a hosted service at mcp.illinihunt.org
No installation needed: Canvas MCP is available as a hosted service, so educators can start using it immediately without any local setup.
Built for the AI Agent Ecosystem
Canvas MCP follows the Model Context Protocol (MCP) standard, making it compatible with any MCP-aware AI agent. This means as new AI coding tools emerge, they can immediately leverage Canvas MCP without any modifications.
Get Started
Canvas MCP is available now on GitHub. Visit the project page for full documentation and setup instructions.