# Submission Evaluator

> Evaluates student submissions for practical assignments, machine exams, and projects using an evidence-first workflow. Use when an instructor needs robust grading from a subject, rubric, reference materials, tests, and student files, especially in programming and systems courses.

- Skill: `alainlebret/submission-evaluator` (Agent Skill)
- Install (CLI): `npx skillmds@latest add alainlebret/submission-evaluator`
- Raw SKILL.md: https://api.skillmd.com/api/skills/alainlebret/submission-evaluator/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Author: alainlebret (https://skillmd.com/u/alainlebret)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/alainlebret/submission-evaluator

---


# Submission Evaluator

## Goal
Grade submissions through structured evidence, not impressionistic judgment.

## Mandatory workflow
1. Read the subject.
2. Read the rubric.
3. Read the reference materials.
4. Collect observable evidence.
5. Compute criterion scores.
6. Generate the final report.

## Required artifacts
When grading many submissions, create:
- `evidence.json` or per-submission evidence files;
- a summary of criterion scores;
- a final report in Markdown.

## Evidence policy
Record at least:
- expected files;
- present files;
- build status;
- test results;
- interface conformity;
- major defects;
- strengths;
- awarded points per criterion.

## Rules
- Never assign final grades before evidence collection.
- If build fails, analyze sources anyway.
- Give partial credit only when supported by evidence.
- Keep student identity distinct from path anomalies or symlink cases.