canvas-mcp

Canvas MCP Server

MCP server for Canvas LMS with up to 126 tools and 8 agent skills. Designed for Claude Desktop, Cursor, Codex, Windsurf, and 40+ other agents; setup and capabilities vary by client.

Quick Start

1. Install the MCP server (everything else, including the skills, needs it running):

• Claude Desktop: download canvas-mcp.mcpb from the latest release and double-click it. It prompts for your Canvas URL and token; no terminal needed. Details.
• Cursor, Zed, Windsurf, Continue, Claude Code and other clients: pip install into a virtualenv, add your token to .env, and point your client at the canvas-mcp-server binary. Local Installation has the per-client config blocks.

2. Verify: canvas-mcp-server --test should report a successful Canvas connection. Then restart your client.

3. Optional: add the workflow skills. These teach your agent the multi-step recipes (weekly plan, morning check, bulk grading, peer review, course QC) on top of the tools:

npx skills add vishalsachdev/canvas-mcp

See Agent Skills for the list. If your agent is Claude Code, the same recipes are also available as slash commands.

For AI Agents

Canvas MCP provides up to 126 tools for interacting with Canvas LMS; the default profile registers fewer, and optional feature-gated tools can raise the total to 126. Tools are organized by user type:

Skills: canvas-course-qc (pre-semester audit), canvas-accessibility-auditor (WCAG-oriented review), canvas-course-builder (scaffold courses from specs/templates).

Decision tree: Simple query → MCP tools. Batch grading (10+) → bulk_grade_submissions. Complex bulk (30+) → execute_typescript.

Quick Reference

Course identifiers: Canvas ID (12345), course code (badm_350_120251_246794), or SIS ID

Cannot do: Create/delete courses, modify course settings, access other users' data

Rate limits: ~700 requests/10 min. Use max_concurrent=5 for bulk operations.

Full documentation: AGENTS.md | tools/TOOL_MANIFEST.json | tools/README.md

Overview

The Canvas MCP Server bridges the gap between AI assistants and Canvas Learning Management System, providing role-specific workflows for students, educators, learning designers, and developers. Built on the Model Context Protocol (MCP), it is designed for MCP-compatible clients; setup and supported capabilities vary by client.

Latest Release: v1.14.0

Released: October 2026 | Full Changelog | All Releases

A student-tools and correctness release, most of it contributed by @KKazuhaK. Students get read-only tools for their groups, grades, quizzes, calendar and a cross-course "what's new" feed, plus Inbox messaging that is off unless the operator enables it. Instructors get guarded body edits and complete reads.

• Student tools (student profile): list_my_groups, get_my_assignment_scores and calculate_grade_scenarios (what-if grades the way Canvas computes them), list_quizzes and get_quiz_details (never start an attempt), list_calendar_events and list_planner_notes, list_my_announcements and get_my_activity_stream. Student writes register only when named in STUDENT_WRITE_TOOLS and remain subject to ALLOWED_WRITE_TOOLS. Messaging (send_message, reply_to_conversation) previews first and reaches at most 5 individual people. Planner-note updates/deletes and personal-calendar deletes also require a preview and confirmation token. create_planner_note, mark_planner_item_complete, and create_personal_calendar_event write immediately on the first valid call; they have no preview step. Completing course content can also change module progression. Get the user's approval before invoking these immediate writes (#465 to #476)
• Guarded body edits: edit_page_content, update_assignment, update_discussion_topic and update_syllabus take expect_updated_at or expect_body_sha256, find/replace and require, and report a write as confirmed only when the read-back proves it (#437)
• Discussions inside group spaces via group_id and list_group_discussion_topics (#433, thanks @papatistos); anonymous discussions read through GraphQL (#439, thanks @papatistos) and refused for updates (#446)
• Complete reads: full-content tools return whole text, previews say they are previews (#458); course codes with spaces, course names and SIS IDs resolve (#457); opt-in raw_dates (#438)
• Behavior change: assign_peer_review never posts a placeholder submission and needs a numeric reviewee_id (#436)

Also fixed: graded paper and no-submission assignments are no longer reported overdue (#423, thanks @lindsay-cheng). The suite runs on Windows in CI (#456, with UTF-8 groundwork by @bruchris in #445). Patched pyjwt, urllib3 and multidict.