canvas-mcp-tool

Exposes Canvas Learning Management System REST API endpoints through the Model Context Protocol (MCP) server interface, enabling Claude and other MCP-compatible clients to authenticate with Canvas instances and execute API calls without direct HTTP handling. Uses MCP's tool-calling schema to map Canvas API operations (courses, assignments, grades, users) into callable functions with standardized request/response formatting.

Implements read-only access to Canvas course structures, assignments, submissions, and metadata through MCP tool functions that query Canvas REST endpoints (/api/v1/courses, /api/v1/courses/:id/assignments, /api/v1/courses/:id/submissions). Returns structured JSON containing course hierarchies, assignment rubrics, due dates, submission status, and student enrollment data with pagination support for large datasets.

Provides write access to Canvas grading operations through MCP tool functions that call Canvas PUT/POST endpoints (/api/v1/courses/:id/assignments/:id/submissions/:id, /api/v1/courses/:id/assignments/:id/submissions/:id/grade). Supports posting grades, adding comments to submissions, updating submission status, and bulk grading operations with validation against assignment rubrics and point scales.

Retrieves Canvas user profiles, enrollment records, and role information through MCP tool functions calling Canvas endpoints (/api/v1/courses/:id/enrollments, /api/v1/users/:id, /api/v1/accounts/:id/users). Returns structured user data including names, email addresses, enrollment status, roles (student/instructor/ta), and course sections with filtering by enrollment type and status.

Implements the MCP server runtime that handles client connections, tool registration, and request routing. Uses Node.js MCP SDK to expose Canvas operations as standardized MCP tools with JSON schema definitions, manages authentication token storage (environment variables or config files), and handles server startup/shutdown with error logging and connection state management.

Implements error handling for Canvas API responses with mapping of HTTP status codes to user-friendly error messages, request validation against Canvas API constraints (e.g., grade ranges, required fields), and retry logic for transient failures. Catches Canvas-specific errors (invalid course_id, permission denied, rate limiting) and translates them into MCP error responses with diagnostic context.