Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Cut AI tells from any writing. Must always apply.
.claude/skills/kunanonj-cursor-plugin-pstack-unslop/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 116% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 177% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 123% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 192% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 129% | 0% |
Edit text to remove AI patterns and add human voice.
Removing patterns is half the job. Sterile, voiceless writing is just as obvious.
.toSQL() returns the exact string sent to the database", "a column rename fails the build". Ask what the sentence tells the reader to do or know, then write that. If you can't restate it as a concrete instruction, fact, or number, cut it.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-11 | pass→pass | 3,954 | 3,401 | -14% | 1 | 1 | 0% | 758 | 2,064 | +172% | 0 | 0 | — |
case-01 | pass→pass | 3,407 | 2,104 | -38% | 1 | 1 | 0% | 541 | 1,875 | +247% | 0 | 0 | — |
case-02 | fail→pass | 6,064 | 4,234 | -30% | 1 | 1 | 0% | 1,004 | 2,166 | +116% | 0 | 0 | — |
case-03 | pass→pass | 4,578 | 5,667 | +24% | 1 | 1 | 0% | 730 | 2,469 | +238% | 0 | 0 | — |
case-04 | fail→fail | 5,741 | 4,654 | -19% | 1 | 1 | 0% | 1,081 | 2,435 | +125% | 0 | 0 | — |
case-05 | fail→pass | 4,175 | 3,851 | -8% | 1 | 1 | 0% | 769 | 2,132 | +177% | 0 | 0 | — |
case-06 | pass→pass | 6,166 | 3,312 | -46% | 1 | 1 | 0% | 1,203 | 2,128 | +77% | 0 | 0 | — |
case-07 | pass→pass | 7,506 | 6,312 | -16% | 1 | 1 | 0% | 1,448 | 2,567 | +77% | 0 | 0 | — |
case-08 | fail→pass | 5,052 | 2,626 | -48% | 1 | 1 | 0% | 883 | 1,966 | +123% | 0 | 0 | — |
case-09 | pass→pass | 4,973 | 3,654 | -27% | 1 | 1 | 0% | 924 | 2,094 | +127% | 0 | 0 | — |
case-10 | fail→pass | 3,996 | 3,458 | -13% | 1 | 1 | 0% | 724 | 2,112 | +192% | 0 | 0 | — |
case-12 | pass→pass | 5,224 | 4,349 | -17% | 1 | 1 | 0% | 915 | 2,340 | +156% | 0 | 0 | — |
case-13 | fail→pass | 5,272 | 2,972 | -44% | 1 | 1 | 0% | 858 | 1,968 | +129% | 0 | 0 | — |
case-14 | fail→pass | 5,323 | 4,199 | -21% | 1 | 1 | 0% | 1,083 | 2,335 | +116% | 0 | 0 | — |
case-15 | fail→pass | 8,283 | 2,390 | -71% | 1 | 1 | 0% | 1,485 | 1,886 | +27% | 0 | 0 | — |
case-16 | pass→pass | 6,767 | 4,309 | -36% | 1 | 1 | 0% | 1,232 | 2,131 | +73% | 0 | 0 | — |
case-17 | pass→pass | 7,899 | 4,325 | -45% | 1 | 1 | 0% | 1,315 | 2,305 | +75% | 0 | 0 | — |
case-18 | pass→pass | 7,210 | 4,487 | -38% | 1 | 1 | 0% | 1,241 | 2,208 | +78% | 0 | 0 | — |
case-19 | pass→fail | 4,104 | 3,189 | -22% | 1 | 1 | 0% | 684 | 2,025 | +196% | 0 | 0 | — |
case-20 | pass→pass | 4,612 | 5,298 | +15% | 1 | 1 | 0% | 940 | 2,576 | +174% | 0 | 0 | — |
case-21 | pass→pass | 5,938 | 5,093 | -14% | 1 | 1 | 0% | 1,117 | 2,402 | +115% | 0 | 0 | — |
case-22 | pass→pass | 3,614 | 3,766 | +4% | 1 | 1 | 0% | 777 | 2,207 | +184% | 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 +27 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is 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.