Canvas MCP Server

Instructions

    Args:
        course_identifier: The Canvas course code (e.g., badm_554_120251_246794) or ID
        assignment_id: The Canvas assignment ID
        reviewer_id: The Canvas user ID of the student who will do the review
        reviewee_id: The Canvas user ID of the student whose submission will be reviewed
    

Implementation Reference

• src/canvas_mcp/tools/assignments.py:92-166 (handler)

  The core handler function for the 'assign_peer_review' MCP tool. Decorated with @mcp.tool() for automatic registration and schema generation. Implements logic to assign peer reviews in Canvas by ensuring reviewee submission exists (creating placeholder if needed) and posting to Canvas peer_reviews API endpoint.

  @mcp.tool()
  async def assign_peer_review(course_identifier: str, assignment_id: str, reviewer_id: str, reviewee_id: str) -> str:
      """Manually assign a peer review to a student for a specific assignment.
  
      Args:
          course_identifier: The Canvas course code (e.g., badm_554_120251_246794) or ID
          assignment_id: The Canvas assignment ID
          reviewer_id: The Canvas user ID of the student who will do the review
          reviewee_id: The Canvas user ID of the student whose submission will be reviewed
      """
      course_id = await get_course_id(course_identifier)
  
      # First, we need to get the submission ID for the reviewee
      submissions = await make_canvas_request(
          "get",
          f"/courses/{course_id}/assignments/{assignment_id}/submissions",
          params={"per_page": 100}
      )
  
      if "error" in submissions:
          return f"Error fetching submissions: {submissions['error']}"
  
      # Find the submission for the reviewee
      reviewee_submission = None
      for submission in submissions:
          if str(submission.get("user_id")) == str(reviewee_id):
              reviewee_submission = submission
              break
  
      # If no submission exists, we need to create a placeholder submission
      if not reviewee_submission:
          # Create a placeholder submission for the reviewee
          placeholder_data = {
              "submission": {
                  "user_id": reviewee_id,
                  "submission_type": "online_text_entry",
                  "body": "Placeholder submission for peer review"
              }
          }
  
          reviewee_submission = await make_canvas_request(
              "post",
              f"/courses/{course_id}/assignments/{assignment_id}/submissions",
              data=placeholder_data
          )
  
          if "error" in reviewee_submission:
              return f"Error creating placeholder submission: {reviewee_submission['error']}"
  
      # Now assign the peer review using the submission ID
      submission_id = reviewee_submission.get("id")
  
      # Data for the peer review assignment
      data = {
          "user_id": reviewer_id  # The user who will do the review
      }
  
      # Make the API request to create the peer review
      response = await make_canvas_request(
          "post",
          f"/courses/{course_id}/assignments/{assignment_id}/submissions/{submission_id}/peer_reviews",
          data=data
      )
  
      if "error" in response:
          return f"Error assigning peer review: {response['error']}"
  
      # Try to get the course code for display
      course_display = await get_course_code(course_id) or course_identifier
  
      return f"Successfully assigned peer review in course {course_display}:\n" + \
             f"Assignment ID: {assignment_id}\n" + \
             f"Reviewer ID: {reviewer_id}\n" + \
             f"Reviewee ID: {reviewee_id}\n" + \
             f"Submission ID: {submission_id}"

• src/canvas_mcp/server.py:48-48 (registration)

  Top-level registration of assignment tools (including 'assign_peer_review') by calling register_assignment_tools(mcp) in the main server setup.

  register_assignment_tools(mcp)