| name | feedback-write |
| description | Write actionable feedback on a placeholder student response compared to a rubric. Use this atom when a workflow needs to model what good feedback looks like for a given rubric and response. Do NOT use with real student data — placeholders only. Do NOT use for grading. |
feedback-write
Generates model feedback on a placeholder response evaluated against a rubric. Shows teachers what actionable, growth-oriented feedback looks like. No real student data ever.
Input
{
"rubric": {"criteria": [{"criterion": "Text evidence", "levels": {"Meets": "Cites 2+ relevant quotes", "Below": "No evidence cited"}}]},
"placeholder_response": "[Student Name] wrote: The character was sad because of the events.",
"grade": "7"
}
Output
{
"tool": "feedback-write",
"feedback": "You identified the character emotion — nice start! To move from Approaching to Meets, go back to paragraph 3 and find a specific quote that shows WHY the character feels this way. Try starting with: According to the text...",
"rubric_level_matched": "Approaching",
"growth_target": "Meets",
"human_review_required": true
}
Do NOT use this atom for
- Using real student data (placeholders only)
- Assigning grades or scores (this models feedback, not grading)
- Generating rubrics (use atom-rubric-build)
Pipeline note
Follows references/method.md at the Generation step (feedback modeling). Output conforms to protocol-layer/metadata-schema.md. human_review_required: true — feedback tone and accuracy must be teacher-reviewed.
