Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Version 2.35.0 | PRD to Production | Zero Human Intervention > Research-enhanced: OpenAI SDK, DeepMind, Anthropic, AWS Bedrock, Agent SDK, HN Production (2025)
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 220% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 433% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 209% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 294% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 372% | 0% |
> Version 2.35.0 | PRD to Production | Zero Human Intervention > Research-enhanced: OpenAI SDK, DeepMind, Anthropic, AWS Bedrock, Agent SDK, HN Production (2025)
Read the detailed guide before executing this skill. It retains the complete procedure and reference material. Treat its safety, prerequisites, and validation requirements as mandatory. For focused work, load the relevant sections; for end-to-end work, read the guide completely.
bash# Launch with autonomous permissions claude
This skill is applicable to execute the workflow or actions described in the overview.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→fail | 4,904 | 5,093 | +4% | 1 | 1 | 0% | 222 | 7,236 | +3159% | 0 | 0 | — |
case-01 | fail→fail | 4,049 | 6,364 | +57% | 1 | 1 | 0% | 247 | 7,326 | +2866% | 0 | 0 | — |
case-02 | fail→fail | 6,447 | 5,012 | -22% | 1 | 1 | 0% | 1,027 | 7,174 | +599% | 0 | 0 | — |
case-04 | fail→pass | 23,782 | 9,596 | -60% | 1 | 1 | 0% | 2,735 | 8,758 | +220% | 0 | 0 | — |
case-05 | fail→pass | 7,345 | 5,167 | -30% | 1 | 1 | 0% | 1,475 | 7,858 | +433% | 0 | 0 | — |
case-06 | fail→pass | 13,708 | 10,486 | -24% | 1 | 1 | 0% | 2,844 | 8,795 | +209% | 0 | 0 | — |
case-07 | pass→pass | 10,317 | 4,243 | -59% | 1 | 1 | 0% | 1,449 | 7,693 | +431% | 0 | 0 | — |
case-08 | pass→pass | 12,979 | 6,525 | -50% | 1 | 1 | 0% | 1,758 | 7,968 | +353% | 0 | 0 | — |
case-09 | fail→fail | 5,989 | 11,850 | +98% | 1 | 1 | 0% | 909 | 7,727 | +750% | 0 | 0 | — |
case-10 | fail→pass | 12,613 | 4,393 | -65% | 1 | 1 | 0% | 1,911 | 7,528 | +294% | 0 | 0 | — |
case-11 | fail→pass | 10,296 | 4,631 | -55% | 1 | 1 | 0% | 1,606 | 7,583 | +372% | 0 | 0 | — |
case-12 | fail→pass | 3,508 | 6,214 | +77% | 1 | 1 | 0% | 467 | 8,021 | +1618% | 0 | 0 | — |
case-13 | fail→pass | 12,231 | 7,009 | -43% | 1 | 1 | 0% | 1,964 | 8,074 | +311% | 0 | 0 | — |
case-14 | pass→pass | 12,486 | 5,385 | -57% | 1 | 1 | 0% | 2,034 | 7,731 | +280% | 0 | 0 | — |
case-15 | fail→pass | 14,937 | 13,189 | -12% | 1 | 1 | 0% | 2,470 | 9,073 | +267% | 0 | 0 | — |
case-16 | pass→pass | 9,712 | 5,805 | -40% | 1 | 1 | 0% | 1,482 | 7,872 | +431% | 0 | 0 | — |
case-17 | pass→pass | 12,987 | 8,311 | -36% | 1 | 1 | 0% | 2,024 | 8,131 | +302% | 0 | 0 | — |
case-18 | pass→pass | 12,963 | 7,602 | -41% | 1 | 1 | 0% | 2,113 | 8,183 | +287% | 0 | 0 | — |
case-19 | pass→pass | 12,679 | 6,594 | -48% | 1 | 1 | 0% | 2,057 | 8,044 | +291% | 0 | 0 | — |
case-20 | pass→pass | 20,695 | 21,733 | +5% | 1 | 1 | 0% | 3,491 | 10,862 | +211% | 0 | 0 | — |
case-21 | fail→fail | 4,128 | 5,062 | +23% | 1 | 1 | 0% | 627 | 7,196 | +1048% | 0 | 0 | — |
case-22 | pass→pass | 3,424 | 3,087 | -10% | 1 | 1 | 0% | 485 | 7,385 | +1423% | 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, and 17 counted toward the lift figure. The other 5 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +36 percentage points is the difference between those two pass rates over the 17 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
The publisher has shipped newer versions since this run, so these numbers describe v2, not the version currently listed.
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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 7/28/2026 | +40% |
Other measured skills in the registry, with their headline benchmark lift.