Canvas Grading Pass

Work through an assignment's ungraded submissions one at a time — viewing the student's work, applying the rubric, entering a score, and adding a comment — with a confirmation gate before each write operation.

Prerequisites

• Canvas MCP server must be running and connected.
• You must have instructor or TA role in the target course.
• The assignment must accept online submissions (file upload, text entry, URL, etc.).

Steps

1. Select the Course and Assignment

Ask the instructor which course and assignment to grade.

If unclear, call list_courses to let them choose a course, then call list_assignments (with the course ID) to select an assignment. Prefer assignments with workflow_state=published and a past due date.

2. Load the Rubric (if attached)

Call list_rubrics for the course to see available rubrics. If the assignment has an associated rubric (check rubric_id in the assignment object from get_assignment), call get_rubric with that rubric ID to retrieve criteria, rating descriptions, and point values.

Display the rubric criteria to the instructor before grading begins so they have it in view throughout.

3. Retrieve Ungraded Submissions

Call list_submissions for the course and assignment. Filter to submissions where workflow_state is submitted (not graded). Present the count:

"Found [n] ungraded submissions for [Assignment]. Grading one at a time."

For assignments with more than 50 students, note: pace your session — Canvas rate-limits write operations. Take a short break between batches of 20–25 submissions if you encounter slowdowns.

4. Grade Each Submission (one at a time)

Repeat steps 4a–4d for each ungraded submission. Do not batch or skip the confirmation gate.

#### 4a. View the Submission

Call get_submission with the course ID, assignment ID, and student user ID. Display:

• Student name and user ID
• Submission date and type
• Submission body, URL, or attachment reference
• Any existing score or comments

Check get_rubric_assessment (with course ID, rubric ID, and submission ID) to see if a partial assessment already exists.

#### 4b. Propose a Score or Rubric Assessment

Based on the rubric (if present) and the instructor's judgment, draft:

• A rubric assessment: ratings per criterion with point values
• Or a raw numeric score

Present the proposed grade and ask the instructor:

"Proposed score: [n]/[max] pts Rubric: [criterion] → [rating] ([pts]) Add comment: [draft comment] Confirm to submit? (yes / edit / skip)"

Do not proceed until the instructor confirms.

#### 4c. Submit Rubric Assessment (if rubric attached)

After confirmation, call submit_rubric_assessment with:

• course_id, rubric_id, rubric_association_id (from the rubric object)
• Assessment data: criterion ID → rating ID + points for each criterion
• graded_anonymously: false unless instructor specifies otherwise

#### 4d. Grade and/or Comment

If no rubric is attached (or in addition to the rubric assessment), call grade_submission with the numeric score.

If the instructor provided a comment, call comment_on_submission with the comment text.

Report: "✓ [Student name] — [score]/[max] pts — comment posted."

5. End-of-Session Summary

After all submissions are reviewed (or the instructor stops the session), present:

Grading Session Complete — [Assignment]

Graded this session:  [n]
Skipped:              [n]
Remaining ungraded:   [n]

Ask if the instructor wants to continue with any skipped submissions or return later.

Output Format

Grading Pass — [Course] › [Assignment]
[n] ungraded  |  [max pts] pts  |  Rubric: [attached / none]

--- Submission 1 of [n] ---
Student:   [Name]  (ID: [id])
Submitted: [datetime]
Type:      [online_text_entry / file / url]

[submission preview or link]

Rubric Assessment (proposed):
  [Criterion 1]  →  [Rating label]  ([pts] pts)
  [Criterion 2]  →  [Rating label]  ([pts] pts)
  Total: [n]/[max] pts

Draft comment: "[comment text]"

Confirm? (yes / edit score / edit comment / skip)

Notes

• This skill grades one submission at a time — there is no bulk grading mode. Every write requires explicit instructor confirmation.
• grade_submission and submit_rubric_assessment are write operations that modify the Canvas gradebook. They cannot be undone via the MCP server; use the Canvas web UI to correct mistakes.
• For assignments with more than 50 students, pace the session. If you encounter API errors, wait 30–60 seconds and retry the single failed call — do not retry the entire batch.
• Anonymous grading: if the assignment has anonymous_grading: true, student names will not be visible in submission data. Note this to the instructor before starting.
• comment_on_submission posts a visible comment to the student. Draft carefully and confirm before posting.