Assess Submit

Step 2: Load and Validate Assessment File

Read the assessment JSON file using the Read tool. Validate structure and compute statistics.

Display validation summary:

Assessment File Status
======================

File: {assessment_file_path}
Assignment: {assignment_name}
Course: {course_name}

Assessments Overview:
  Total submissions: X
  Reviewed: Y (AI-evaluated)
  Approved: Z (ready to submit)
  Not reviewed: W

Score Statistics (approved only):
  Average: {avg} / 20 pts
  Range: {min} - {max} pts

Step 3: Dry Run Preview

Run the dauber CLI submit command without --confirm to get a dry run:

uv run dauber assess submit {assessment_file_path} {course_id} {assignment_id} --format json

This shows what would be submitted without making changes. Build and display a preview from the assessment file data:

Submission Preview (DRY RUN)
============================

WILL BE SUBMITTED (Z approved):
  User {user_id} - {total_pts}/{points_possible} pts
    - {criterion1}: {pts1}/{max1}
    - {criterion2}: {pts2}/{max2}
    - ...
  (list all criteria from rubric)

WILL BE SKIPPED:
  User {user_id} - not approved (needs review)
  User {user_id} - already graded in Canvas (protected)
  ...

NO SUBMISSION (W students):
  - User {user_id} - no submission to grade
  ...

Step 4: Confirm Submission

Ask user for confirmation:

Ready to submit Z approved assessments to Canvas?

This will:

  - Submit grades for Z students
  - Skip X unapproved assessments
  - Post rubric scores and feedback to Canvas
  - Mark submitted assessments in JSON file

Continue? (Y/n)

Use AskUserQuestion tool with:

• Question: "Submit approved assessments to Canvas gradebook?"
• Options: ["Yes - submit to Canvas", "No - cancel submission"]

Step 5: Submit to Canvas

If user confirms YES:

Run the dauber CLI submit command with --confirm:

uv run dauber assess submit {assessment_file_path} {course_id} {assignment_id} --confirm --format json

Report results:

Submission Complete
===================

Successfully Submitted: X students
  User {user_id} - {total_pts}/{points_possible} pts posted to Canvas
  ...

Skipped:
  Not approved: Y students
  Already graded in Canvas: Z students (protected)
  ...

Failed (errors): W students
  User {user_id} - Error: {error_message}
  ...

Grades Posted to Canvas: X/{total_submissions}
Assessment File Updated

If user cancels:

• Report: "Submission cancelled. No grades posted to Canvas."
• Exit without making changes

Step 6: Update State

Update .claude/state/assessment_state.json with:

{
  "last_assessment_file": "{file_path}",
  "course_id": "{course_id}",
  "assignment_id": "{assignment_id}",
  "last_submission_time": "{timestamp}",
  "total_submitted": X,
  "total_approved": Y,
  "total_skipped": Z
}

Post-Submission Summary

Final Summary
=============

Assignment: {assignment_name}
Course: {course_name}

Grading Complete:
  X students graded in Canvas
  Y students pending approval
  - Z students no submission

Class Performance (submitted grades):
  Average: {avg}/{points_possible} pts ({percentage}%)

  Criterion Averages:
  - {criterion1}: {avg1}/{max1} pts
  - {criterion2}: {avg2}/{max2} pts
  - ...
  (list all criteria from rubric)

Assessment File: {file_path}
  Preserved as audit trail with submission timestamps

Next Steps:
  - Review skipped assessments if any
  - Check Canvas gradebook to verify grades posted
  - Follow up with students who didn't submit

Error Handling

• If no approved assessments: Report warning, ask if user wants to review assessments first
• If submission fails for any student: Report error but continue with others
• If Canvas API error: Report error with details, preserve assessment file
• If assessment file not found: Report error, tell user to run setup/prelim first

Important Notes

• ALWAYS runs dry-run preview first before actual submission
• ONLY submits assessments explicitly marked as "approved"
• Protects existing Canvas grades by default (won't overwrite already-graded)
• Skips unapproved assessments and already-graded submissions
• Preserves assessment file as audit trail
• Updates JSON with submission timestamps
• No grades can be submitted without human approval
• Safe to run multiple times (won't duplicate submissions or overwrite existing grades)

Begin submission process now.