canvas-mcp
Connects Canvas LMS to Claude Desktop, enabling natural-language queries about courses, assignments, grades, and more.
README
canvas-mcp
A Model Context Protocol (MCP) server that connects your Canvas LMS account to Claude Desktop. Ask Claude natural-language questions about your courses, assignments, grades, files, and more -- all powered by the Canvas REST API.
Features
canvas-mcp exposes 15 tools to Claude Desktop:
| Tool | Description |
|---|---|
get_courses |
List all active enrolled courses |
get_assignments |
Assignments for a specific course with submission status |
get_all_assignments |
Every assignment across all courses, sorted by due date |
get_upcoming_assignments |
Assignments due in the next N days (default 7) |
get_calendar_events |
Upcoming calendar events (next 30 days) |
get_submission_status |
Per-assignment submission state (submitted / graded / unsubmitted) |
get_grades |
Current and final grade plus per-assignment scores |
get_course_modules |
Modules with item completion status |
get_announcements |
Recent announcements across all courses |
get_todo_items |
Your Canvas to-do list |
get_course_files |
List files uploaded to a course (PDFs, slides, etc.) |
get_course_pages |
List wiki/content pages in a course |
get_page_content |
Read the full text of a specific course page |
read_course_file |
Download and parse a file -- full PDF text extraction |
get_module_item_content |
Read content of all items in a module (lectures, notes, PDFs) |
Project Structure
canvas-mcp/
├── src/
│ ├── index.ts # MCP server entry point, tool definitions, and dispatch
│ ├── tools.ts # Tool implementation functions
│ └── canvas-client.ts # Canvas API client: HTTP helpers, pagination, type definitions
├── dist/ # Compiled JavaScript (generated by build)
├── .env # Your credentials (not committed)
├── .gitignore
├── package.json
├── tsconfig.json
└── README.md
Prerequisites
- Node.js >= 18
- A Canvas LMS API access token
Getting a Canvas API Token
- Log in to your Canvas instance (e.g.
https://yourschool.instructure.com) - Go to Account > Settings
- Scroll to Approved Integrations
- Click + New Access Token
- Give it a name (e.g. "Claude MCP") and optionally set an expiry
- Copy the token -- you will not be able to see it again
Setup
1. Clone the repository
git clone https://github.com/YOUR_USERNAME/canvas-mcp.git
cd canvas-mcp
2. Install dependencies
npm install
3. Configure environment variables
Create a .env file in the project root:
CANVAS_TOKEN=your_canvas_api_token_here
CANVAS_DOMAIN=yourschool.instructure.com
CANVAS_DOMAIN should be just the hostname (no https:// prefix).
4. Build
npm run build
This compiles the TypeScript source to dist/.
5. Verify (optional)
node dist/index.js
The server communicates over stdio using the MCP protocol. It will start and wait for input. Press Ctrl+C to exit. If environment variables are missing, you will see a clear error immediately.
Connecting to Claude Desktop
Add the server to your Claude Desktop configuration file.
macOS
Edit ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"canvas": {
"command": "node",
"args": ["/absolute/path/to/canvas-mcp/dist/index.js"],
"env": {
"CANVAS_TOKEN": "your_canvas_token_here",
"CANVAS_DOMAIN": "yourschool.instructure.com"
}
}
}
}
Windows
Edit %APPDATA%\Claude\claude_desktop_config.json:
{
"mcpServers": {
"canvas": {
"command": "node",
"args": ["C:\\Users\\YOU\\path\\to\\canvas-mcp\\dist\\index.js"],
"env": {
"CANVAS_TOKEN": "your_canvas_token_here",
"CANVAS_DOMAIN": "yourschool.instructure.com"
}
}
}
}
Replace the path with the absolute path to your compiled dist/index.js. After saving, restart Claude Desktop.
Example Prompts
Once connected, try these in Claude Desktop:
Assignment planning
- "What assignments do I have due this week?"
- "Show me everything due in the next 14 days."
- "Build me a study schedule for the next two weeks based on my due dates."
Submission tracking
- "Am I missing any submissions?"
- "Which assignments haven't I submitted yet in course 12345?"
- "Show me all overdue assignments."
Grades
- "What's my current grade in my Biology course?"
- "Show me all my graded assignments and scores."
- "Which course am I doing worst in?"
Course content
- "What courses am I enrolled in this semester?"
- "What was covered in this week's lectures for course 12345?"
- "Read the syllabus PDF from my math course."
- "What modules have I completed in course 12345?"
Comprehensive
- "Give me a full academic status report -- courses, grades, upcoming work, and missing submissions."
- "I have a free weekend. What should I prioritize based on upcoming deadlines and my current grades?"
Development
# Run in dev mode without building (uses ts-node)
npm run dev
# Rebuild after making changes
npm run build
Troubleshooting
| Problem | Solution |
|---|---|
Missing required environment variables |
Check your .env file or the env block in claude_desktop_config.json |
Canvas API error 401 |
Your token is invalid or expired -- generate a new one |
Canvas API error 403 |
You do not have permission to access that resource |
| No active courses found | Your enrollment state may differ -- verify in Canvas directly |
| Claude does not see the server | Verify the absolute path in your config and restart Claude Desktop |
License
This project does not currently include a license file.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。