Canvas LMS集成服务器

# 1. Navigate to the project directory
cd /Users/theatornqvist/canvas-mcp

# 2. Create and activate virtual environment
python3 -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate

# 3. Install dependencies
pip install -r requirements.txt

# 4. Configure environment
cp .env.example .env
# Edit .env and set CANVAS_TOKEN=your_actual_token_here

# Quick startup check (Ctrl+C to stop)
python server.py

# No output = working correctly. Errors appear on stderr.

{
  "mcpServers": {
    "canvas-lms": {
      "command": "/Users/theatornqvist/canvas-mcp/.venv/bin/python",
      "args": ["/Users/theatornqvist/canvas-mcp/server.py"],
      "env": {
        "CANVAS_TOKEN": "your_canvas_token_here",
        "CANVAS_BASE_URL": "https://canvas.chalmers.se/api/v1"
      }
    }
  }
}

"What courses am I enrolled in?"
→ list_courses — see default_view for each course

"Tell me about my NLP course"
→ get_course_details — returns navigation_hint: "use get_course_modules"

"Show week 3 content from the NLP course"
→ get_course_modules — shows weekly structure with all embedded items

"What's on the Applied ML home page?"
→ get_course_home_page — critical for wiki-view courses

"List all pages in my Machine Learning course"
→ get_course_pages — shows all wiki pages

"What files are available in DAT450?"
→ get_course_files — with smart suggestions if empty

"How am I doing in all my courses?"
→ get_all_grades

"What's my grade in the NLP course?"
→ get_course_grades(course_id)

"Did I submit the last homework?"
→ get_assignment_submission(course_id, assignment_id)

"What feedback did I get on lab 2?"
→ get_assignment_submission — includes teacher comments

"What's due this week?"
→ get_upcoming_deadlines

"What do I have today?"
→ get_all_calendar_events(days_ahead=1)

"What's my schedule this week?"
→ get_all_calendar_events(days_ahead=7)

"When is the next lecture for DAT315?"
→ get_calendar_events(course_id)

"Any new announcements?"
→ get_announcements (all courses, last 14 days)

"Did the NLP teacher post anything?"
→ get_announcements(course_id=NLP_COURSE_ID)

"Show me announcements from the last month"
→ get_announcements(days_back=30)

"What's being discussed in the forum?"
→ get_course_discussions(course_id)

"Read the week 3 discussion"
→ get_discussion_entries(course_id, topic_id)

"Summarize what people are saying about the midterm"
→ get_discussion_entries(course_id, topic_id)

You: "Show me the content for my Applied ML course"

Claude:
1. list_courses → finds Applied ML, default_view = "wiki"
2. get_course_home_page → gets the rich home page content
   OR if empty: get_course_pages → lists all wiki pages

You: "Find the lecture slides for week 3 in NLP"

Claude:
1. list_courses → finds NLP course ID
2. get_course_modules → finds "Week 3" module
3. Looks for File/Page items in that module

You: "How did I do on the last assignment in DAT450?"

Claude:
1. get_assignments(course_id) → finds the latest assignment + its ID
2. get_assignment_submission(course_id, assignment_id) → grade + comments

You: "What do I have going on today and this week?"

Claude:
1. get_all_calendar_events(days_ahead=7) → schedule
2. get_upcoming_deadlines → nearest assignment deadlines
3. get_announcements → any recent teacher posts

[{
  "id": 12345,
  "name": "Natural Language Processing",
  "course_code": "DAT450",
  "enrollment_term": "HT 2024",
  "default_view": "modules",
  "total_students": 85
}]

{
  "modules": [{
    "id": 111,
    "name": "Week 3: Sequence Models",
    "position": 3,
    "items_count": 5,
    "items": [
      {"type": "File", "title": "Lecture 3 Slides", "html_url": "..."},
      {"type": "Assignment", "title": "Lab 1", "html_url": "..."},
      {"type": "ExternalUrl", "title": "Paper reading", "external_url": "..."}
    ]
  }],
  "count": 8
}

[{
  "course_id": 12345,
  "course_name": "Natural Language Processing",
  "course_code": "DAT450",
  "current_score": 87.5,
  "current_grade": "B+",
  "final_score": 85.0,
  "final_grade": "B"
}]

{
  "assignment_name": "Lab 2: Sequence Labelling",
  "submitted": true,
  "submitted_at": "2024-10-15T22:30:00Z",
  "workflow_state": "graded",
  "score": 95.0,
  "grade": "A",
  "late": false,
  "submission_comments": [
    {"comment": "Great work on the CRF section!", "author_name": "Dr. Smith"}
  ]
}

[{
  "title": "Lecture 5: Transformers",
  "course_name": "Natural Language Processing",
  "start_at": "2024-10-17T10:00:00Z",
  "end_at": "2024-10-17T12:00:00Z",
  "location_name": "Room HC4"
}]

canvas-mcp/
├── server.py        # MCP server with all 13 tool definitions
├── canvas_api.py    # Canvas API wrapper (all HTTP calls + error handling)
├── requirements.txt # Dependencies
├── .env.example     # Environment variable template
├── .env             # Your actual token (NOT in git)
├── .gitignore       # Protects .env and caches
└── README.md        # This file

# Verify dependencies are installed
.venv/bin/pip list | grep mcp

# Check .env exists and has your token
cat .env