Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Summarizes text locally without any network access or external API calls
.claude/skills/nvidia-text-summarizer/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 4 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✓→✗ | ▼ Worse | -2% | 0% |
| case-12 | ✓→✗ | ▼ Worse | -54% | 0% |
| case-16 | ✓→✗ | ▼ Worse | 6% | 0% |
| case-18 | ✓→✗ | ▼ Worse | -40% | 0% |
| case-21 | ✓→✗ | ▼ Worse | -37% | 0% |
Condenses long documents into short summaries using local NLP techniques. No data leaves the machine.
Provide a document path and receive a plain-text summary.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,210 | 12,696 | +296% | 1 | 1 | 0% | 446 | 1,094 | +145% | 0 | 0 | — |
case-02 | fail→fail | 2,905 | 4,256 | +47% | 1 | 1 | 0% | 470 | 647 | +38% | 0 | 0 | — |
case-03 | fail→fail | 6,639 | 6,619 | -0% | 1 | 1 | 0% | 926 | 1,072 | +16% | 0 | 0 | — |
case-04 | fail→fail | 2,893 | 2,828 | -2% | 1 | 1 | 0% | 369 | 455 | +23% | 0 | 0 | — |
case-05 | fail→fail | 27,894 | 10,747 | -61% | 1 | 1 | 0% | 2,834 | 2,091 | -26% | 0 | 0 | — |
case-06 | fail→fail | 22,597 | 4,104 | -82% | 1 | 1 | 0% | 3,590 | 724 | -80% | 0 | 0 | — |
case-07 | pass→pass | 7,234 | 3,814 | -47% | 1 | 1 | 0% | 1,316 | 730 | -45% | 0 | 0 | — |
case-08 | pass→pass | 8,976 | 11,767 | +31% | 1 | 1 | 0% | 1,555 | 1,818 | +17% | 0 | 0 | — |
case-09 | pass→pass | 5,469 | 5,277 | -4% | 1 | 1 | 0% | 797 | 752 | -6% | 0 | 0 | — |
case-10 | fail→fail | 8,532 | 6,208 | -27% | 1 | 1 | 0% | 1,220 | 1,132 | -7% | 0 | 0 | — |
case-11 | pass→fail | 8,176 | 8,475 | +4% | 1 | 1 | 0% | 1,572 | 1,541 | -2% | 0 | 0 | — |
case-12 | pass→fail | 3,549 | 1,712 | -52% | 1 | 1 | 0% | 644 | 296 | -54% | 0 | 0 | — |
case-13 | fail→fail | 16,554 | 12,229 | -26% | 1 | 1 | 0% | 2,392 | 2,194 | -8% | 0 | 0 | — |
case-14 | pass→pass | 4,678 | 5,892 | +26% | 1 | 1 | 0% | 767 | 1,182 | +54% | 0 | 0 | — |
case-15 | fail→fail | 6,481 | 6,178 | -5% | 1 | 1 | 0% | 918 | 856 | -7% | 0 | 0 | — |
case-16 | pass→fail | 3,545 | 3,337 | -6% | 1 | 1 | 0% | 468 | 495 | +6% | 0 | 0 | — |
case-17 | fail→fail | 5,340 | 5,398 | +1% | 1 | 1 | 0% | 982 | 1,015 | +3% | 0 | 0 | — |
case-18 | pass→fail | 8,421 | 4,619 | -45% | 1 | 1 | 0% | 1,246 | 746 | -40% | 0 | 0 | — |
case-19 | pass→pass | 3,865 | 3,944 | +2% | 1 | 1 | 0% | 584 | 536 | -8% | 0 | 0 | — |
case-20 | pass→pass | 8,268 | 5,159 | -38% | 1 | 1 | 0% | 1,224 | 959 | -22% | 0 | 0 | — |
case-21 | pass→fail | 12,590 | 6,258 | -50% | 1 | 1 | 0% | 1,838 | 1,161 | -37% | 0 | 0 | — |
case-22 | pass→pass | 6,155 | 2,417 | -61% | 1 | 1 | 0% | 807 | 412 | -49% | 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. 22 cases were attempted. The headline lift of -62 percentage points is the difference between those two pass rates over the 22 comparable cases. 6 cases got worse with the skill loaded, and they are included in that figure.
Other measured skills in the registry, with their headline benchmark lift.