Canvas MCP Server
A remote MCP server for querying Canvas LMS courses, assignments, and grades. Enables natural language interaction with Canvas data via MCP clients like Claude Desktop.
README
Canvas MCP Server
Read-only Remote MCP server (Streamable HTTP transport) for querying Canvas LMS courses, assignments, and grades.
Setup
-
Install dependencies:
npm install -
Configure Canvas credentials:
cp .env.example .envEdit
.envand add:CANVAS_BASE_URL: Your Canvas instance URL (e.g.,https://yourschool.instructure.com)CANVAS_API_TOKEN: Your Canvas personal access tokenCANVAS_TIMEOUT_MS(optional, default15000): Timeout in milliseconds for outbound Canvas API requests (connect + read)PORT(optional, default8080): HTTP server portBASE_PATH(optional, default/mcp): MCP endpoint pathALLOWED_ORIGINS(optional): Comma-separated list of allowed origins for Origin validation (default:http://localhost:*,http://127.0.0.1:*)MCP_AUTH_TOKEN(optional): If set, requireAuthorization: Bearer <token>on all/mcprequests (both GET and POST)
To get your API token:
- Log into Canvas
- Go to Account → Settings
- Scroll to "Approved Integrations"
- Click "+ New Access Token"
- Set purpose (e.g., "MCP Server"), leave expiry blank
- Copy the token (you won't see it again!)
-
Build the project:
npm run build
Running the Server
Start the HTTP server:
npm run dev
The server will start on http://0.0.0.0:8080 (or your configured PORT) with:
- MCP endpoint:
http://localhost:8080/mcp - Health check:
http://localhost:8080/healthz
Test health check:
curl http://localhost:8080/healthz
Expected response: OK
Note: The MCP endpoint (/mcp) uses Streamable HTTP transport (Server-Sent Events) and should be accessed by MCP clients (like Claude Desktop), not directly via curl.
Security & Ops Features:
- Origin validation protects against unauthorized cross-origin requests
- DNS rebinding protection validates the Host header
- Supports localhost, private IP ranges (192.168.x.x, 10.x.x.x, 172.16-31.x.x), and .local domains (mDNS)
- Configure allowed origins via
ALLOWED_ORIGINSenvironment variable - Outbound Canvas API timeouts via
CANVAS_TIMEOUT_MS(default 15000ms) - Structured per-request logging for every MCP tool call (never logs secrets)
- Optional Bearer auth for
/mcpendpoint usingMCP_AUTH_TOKEN
MCP Auth (optional):
- If
MCP_AUTH_TOKENis set, all/mcprequests must include header:Authorization: Bearer <token> - Auth is enforced after Host/Origin validation
- On missing/invalid token, server returns
401with{ "error": "unauthorized" } - The token is never logged
Docker Deployment (Raspberry Pi / Production)
This project includes Docker support for production deployment on Raspberry Pi or any ARM/x64 system.
Prerequisites
- Docker installed on your system
- Docker Compose installed
Deployment Steps
-
Clone the repository to your Raspberry Pi:
git clone <your-repo-url> cd Canvas_MCP -
Create
.envfile with your Canvas credentials:cp .env.example .env nano .envSet your Canvas credentials:
CANVAS_BASE_URL=https://yourschool.instructure.com CANVAS_API_TOKEN=your_canvas_api_token_here # Optional: outbound Canvas API timeout (ms) CANVAS_TIMEOUT_MS=15000 # Optional: require Bearer token auth for /mcp MCP_AUTH_TOKEN=choose-a-strong-token -
Build and start the container:
docker-compose up -d -
Verify the container is running:
docker-compose ps -
Check logs:
docker-compose logs -f -
Test the server:
curl http://localhost:8080/healthz
Docker Compose Commands
Start the server:
docker-compose up -d
Stop the server:
docker-compose down
Restart the server:
docker-compose restart
View logs:
docker-compose logs -f canvas-mcp
Rebuild after code changes:
docker-compose down
docker-compose build --no-cache
docker-compose up -d
Configuration
The container automatically:
- Restarts unless explicitly stopped (
restart: unless-stopped) - Exposes port 8080
- Includes health checks
- Runs as non-root user for security
To change the port, edit docker-compose.yml:
ports:
- "3000:8080" # External:Internal
Connecting to Claude Desktop
This is a remote MCP server using Streamable HTTP transport. Configure Claude Desktop to connect to the running server.
Add to Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"canvas": {
"url": "http://localhost:8080/mcp"
}
}
}
For Raspberry Pi or remote servers:
{
"mcpServers": {
"canvas": {
"url": "http://raspberrypi.local:8080/mcp"
}
}
}
Or use the IP address:
{
"mcpServers": {
"canvas": {
"url": "http://192.168.1.100:8080/mcp"
}
}
}
Important:
- The server must be running before starting Claude Desktop
- Start with Docker:
docker-compose up -dor locally:npm run dev - Verify the server is running:
curl http://localhost:8080/healthz - For remote connections, ensure port 8080 is accessible (firewall rules, etc.)
Restart Claude Desktop, then try:
- "What Canvas courses am I enrolled in?"
- "Show me assignments for [course name]"
- "What assignments am I missing?"
- "Check my submission status for [assignment name]"
- "What's my current grade in [course name]?"
- "What's due this week?"
- "Show me all overdue assignments"
Current Features (v0.5)
list_courses
List all active Canvas courses with course ID, name, and course code.
list_assignments
List assignments for a specific course with details including:
- Assignment ID, name, due date
- Points possible, submission types
- Submission status (workflow state, submitted date, missing/late flags)
Parameters:
course_id(required): The Canvas course IDinclude_future(optional, defaulttrue): Include locked/future assignments. Best-effort filter based onunlock_atfield; assignments withoutunlock_atare always included.status_filter(optional, default"all"): Filter by submission status:"all": No filtering (returns all assignments)"missing":submission.missing === trueOR (due date passed AND nosubmitted_at)"unsubmitted": No submission object OR nosubmitted_atexists (regardless of due date)"submitted":submitted_atexists ORworkflow_stateissubmitted/graded
Supports pagination via Canvas Link headers.
get_submission_status
Get detailed submission status for a specific assignment with a single API request.
Returns:
assignment_id: The assignment IDname: Assignment nameworkflow_state: Current submission state (e.g., "unsubmitted", "submitted", "graded")submitted_at: Submission timestamp (ISO 8601, null if not submitted)graded_at: Grading timestamp (ISO 8601, null if not graded)score: Numeric score (null if not graded)late: Boolean flag indicating late submissionmissing: Boolean flag indicating missing assignmentexcused: Boolean flag indicating excused assignment
Parameters:
course_id(required): The Canvas course IDassignment_id(required): The Canvas assignment ID
get_course_grades
Get grade summary for a course with graceful degradation. Single API request, read-only, never throws. Handles multiple enrollments by preferring active, most current enrollment.
Returns when grades are available:
{
"course_id": 123456,
"available": true,
"current_score": 87.5,
"current_grade": "B+",
"final_score": 85.0,
"final_grade": "B",
"enrollment_state": "active",
"term_id": 5678,
"course_start_at": "2025-01-15T00:00:00Z",
"course_end_at": "2025-05-15T00:00:00Z",
"last_updated": null
}
Returns when grades not yet posted:
{
"course_id": 123456,
"available": false,
"reason": "no_grades_yet",
"enrollment_state": "active",
"term_id": 5678,
"course_start_at": "2025-01-15T00:00:00Z",
"course_end_at": "2025-05-15T00:00:00Z"
}
Returns when grades are hidden or course not found:
{
"course_id": 123456,
"available": false,
"reason": "hidden_or_unavailable"
}
Parameters:
course_id(required): The Canvas course ID
Behavior:
- Handles multiple enrollments (e.g., retaking a course) by preferring
enrollment_state="active"and the most current enrollment (by course end date or enrollment ID) - Distinguishes between grades not yet posted (
no_grades_yet) and grades hidden/unavailable (hidden_or_unavailable) - Includes metadata for sanity-checking:
enrollment_state,term_id, and course date range - All grade fields (scores, grades) may be
nullif not provided by Canvas - Never throws errors
list_upcoming
List upcoming and/or overdue assignments across all active courses in a single consolidated view. Reuses existing list_courses and list_assignments logic.
Example response:
[
{
"course_id": 123456,
"course_name": "Introduction to Computer Science",
"assignment_id": 789012,
"name": "Homework 5",
"due_at": "2025-01-12T23:59:00Z",
"status": "unsubmitted",
"points_possible": 100
},
{
"course_id": 123456,
"course_name": "Introduction to Computer Science",
"assignment_id": 789013,
"name": "Final Project",
"due_at": "2025-01-20T23:59:00Z",
"status": "submitted",
"points_possible": 200
}
]
Parameters:
days(optional, default14): Number of days to look ahead for upcoming assignmentsinclude_overdue(optional, defaulttrue): Include overdue assignmentscourse_ids(optional): Array of course IDs to filter. If not provided, checks all active courses.
Behavior:
- Returns assignments due within the next N days and optionally overdue assignments
- Sorted by due date ascending (overdue items appear first)
- Status is one of:
"submitted","unsubmitted","missing" - Skips assignments without due dates
- Gracefully handles errors on individual courses (continues with remaining courses)
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.