Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write constructive, specific feedback on student work that motivates and tells the student exactly how to improve. Use when asked to give feedback on a student's work, write grading comments, respond to an essay or assignment, or coach a learner. Produces feedback that names concrete strengths, prioritises the few changes that matter most, and gives an actionable next step — warm in tone, growth-oriented, never just a grade.
.claude/skills/mohitagw15856-student-feedback/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 73% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 59% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 25% | 0% |
Feedback changes learning only when it's specific, prioritised, and actionable. This skill writes comments that tell a student what worked, the one or two things to fix next, and exactly how — in a tone that keeps them motivated.
Given the work (or a description of it) and the level, write the full feedback anyway. If the actual work isn't pasted, give a strong model of well-structured feedback and note it should be grounded in the specific submission. No empty "comment]" placeholders.
Ask for (if not already provided):
2–3 concrete strengths tied to the work ("your thesis in paragraph 1 is arguable and clear"), not "good job."
The 1–3 highest-leverage changes, in order. For each: what to change, why it matters, and a concrete example or model of the better version. Don't list every error — prioritise.
One specific, doable action for the next draft or assignment ("in your next essay, start each paragraph with a claim, then evidence").
Genuine, growth-oriented ("you're close — tightening your evidence will lift this a whole level"), not empty praise.
If a rubric was given, map the feedback to its criteria.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | pass→fail | 15,523 | 16,763 | +8% | 1 | 1 | 0% | 1,775 | 2,619 | +48% | 0 | 0 | — |
case-01 | fail→fail | 19,934 | 16,025 | -20% | 1 | 1 | 0% | 2,005 | 2,214 | +10% | 0 | 0 | — |
case-02 | fail→fail | 14,056 | 14,996 | +7% | 1 | 1 | 0% | 1,901 | 2,258 | +19% | 0 | 0 | — |
case-04 | pass→pass | 21,899 | 17,631 | -19% | 1 | 1 | 0% | 2,850 | 2,618 | -8% | 0 | 0 | — |
case-05 | fail→pass | 15,951 | 20,424 | +28% | 1 | 1 | 0% | 2,148 | 3,725 | +73% | 0 | 0 | — |
case-06 | fail→fail | 19,496 | 16,682 | -14% | 1 | 1 | 0% | 2,394 | 2,354 | -2% | 0 | 0 | — |
case-07 | fail→pass | 18,819 | 14,733 | -22% | 1 | 1 | 0% | 1,882 | 2,126 | +13% | 0 | 0 | — |
case-08 | fail→pass | 19,923 | 18,491 | -7% | 1 | 1 | 0% | 2,079 | 2,537 | +22% | 0 | 0 | — |
case-09 | pass→pass | 14,880 | 17,654 | +19% | 1 | 1 | 0% | 1,593 | 2,134 | +34% | 0 | 0 | — |
case-10 | fail→fail | 18,373 | 13,836 | -25% | 1 | 1 | 0% | 1,719 | 1,962 | +14% | 0 | 0 | — |
case-11 | fail→pass | 14,620 | 17,514 | +20% | 1 | 1 | 0% | 1,555 | 2,476 | +59% | 0 | 0 | — |
case-12 | pass→pass | 12,098 | 14,733 | +22% | 1 | 1 | 0% | 1,860 | 2,072 | +11% | 0 | 0 | — |
case-13 | fail→pass | 14,123 | 14,853 | +5% | 1 | 1 | 0% | 2,057 | 2,578 | +25% | 0 | 0 | — |
case-14 | pass→pass | 14,972 | 10,625 | -29% | 1 | 1 | 0% | 1,323 | 2,111 | +60% | 0 | 0 | — |
case-15 | pass→pass | 14,906 | 15,214 | +2% | 1 | 1 | 0% | 1,461 | 2,024 | +39% | 0 | 0 | — |
case-16 | pass→fail | 10,110 | 13,265 | +31% | 1 | 1 | 0% | 1,561 | 1,821 | +17% | 0 | 0 | — |
case-17 | pass→pass | 19,768 | 19,215 | -3% | 1 | 1 | 0% | 2,017 | 2,400 | +19% | 0 | 0 | — |
case-18 | fail→pass | 17,007 | 19,708 | +16% | 1 | 1 | 0% | 2,094 | 2,062 | -2% | 0 | 0 | — |
case-19 | fail→pass | 15,848 | 11,461 | -28% | 1 | 1 | 0% | 1,558 | 2,177 | +40% | 0 | 0 | — |
case-20 | pass→pass | 11,331 | 10,502 | -7% | 1 | 1 | 0% | 1,672 | 2,006 | +20% | 0 | 0 | — |
case-21 | fail→pass | 11,129 | 10,859 | -2% | 1 | 1 | 0% | 1,655 | 2,261 | +37% | 0 | 0 | — |
case-22 | pass→pass | 11,392 | 11,145 | -2% | 1 | 1 | 0% | 1,851 | 2,487 | +34% | 0 | 0 | — |
case-23 | pass→fail | 15,038 | 16,740 | +11% | 1 | 1 | 0% | 2,510 | 3,069 | +22% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +22 percentage points is the difference between those two pass rates over the 23 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.