Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Fully automated review pipeline: CEO, Design, and Eng reviews with auto-decisions using 6 principles (completeness, boil lakes, pragmatic, DRY, explicit over clever, bias toward action). Only pauses for premise confirmation. Classifies decisions as Mechanical (silent) or Taste (surfaced at final gate).
.claude/skills/paperclipai-autoplan/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 94% | 32 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -41% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -48% | 0% |
name: autoplan description: > Fully automated review pipeline: CEO, Design, and Eng reviews with auto-decisions using 6 principles (completeness, boil lakes, pragmatic, DRY, explicit over clever, bias toward action). Only pauses for premise confirmation. Classifies decisions as Mechanical (silent) or Taste (surfaced at final gate). metadata: sources:
repo: garrytan/gstack path: autoplan/SKILL.md commit: f4bbfaa5bdfd2d6ce59541c2145432febde57fed attribution: Garry Tan license: MIT usage: referenced
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 16,774 | 7,556 | -55% | 1 | 1 | 0% | 3,247 | 1,402 | -57% | 0 | 0 | — |
case-02 | pass→pass | 5,576 | 4,263 | -24% | 1 | 1 | 0% | 978 | 855 | -13% | 0 | 0 | — |
case-03 | fail→pass | 13,699 | 7,885 | -42% | 1 | 1 | 0% | 2,387 | 1,398 | -41% | 0 | 0 | — |
case-04 | pass→pass | 6,453 | 4,373 | -32% | 1 | 1 | 0% | 1,015 | 849 | -16% | 0 | 0 | — |
case-05 | fail→pass | 10,177 | 7,441 | -27% | 1 | 1 | 0% | 1,524 | 1,284 | -16% | 0 | 0 | — |
case-06 | pass→pass | 10,692 | 5,431 | -49% | 1 | 1 | 0% | 1,739 | 1,034 | -41% | 0 | 0 | — |
case-07 | pass→pass | 11,778 | 7,370 | -37% | 1 | 1 | 0% | 1,821 | 1,392 | -24% | 0 | 0 | — |
case-08 | pass→pass | 8,256 | 6,507 | -21% | 1 | 1 | 0% | 1,347 | 1,200 | -11% | 0 | 0 | — |
case-09 | pass→pass | 6,957 | 4,168 | -40% | 1 | 1 | 0% | 971 | 831 | -14% | 0 | 0 | — |
case-10 | pass→pass | 10,773 | 8,259 | -23% | 1 | 1 | 0% | 1,847 | 1,499 | -19% | 0 | 0 | — |
case-11 | pass→pass | 7,828 | 5,296 | -32% | 1 | 1 | 0% | 1,281 | 903 | -30% | 0 | 0 | — |
case-12 | fail→pass | 7,447 | 6,171 | -17% | 1 | 1 | 0% | 1,222 | 1,079 | -12% | 0 | 0 | — |
case-13 | pass→pass | 11,226 | 8,190 | -27% | 1 | 1 | 0% | 2,096 | 1,571 | -25% | 0 | 0 | — |
case-14 | fail→pass | 14,373 | 5,093 | -65% | 1 | 1 | 0% | 1,879 | 976 | -48% | 0 | 0 | — |
case-15 | pass→pass | 13,500 | 5,175 | -62% | 1 | 1 | 0% | 1,679 | 969 | -42% | 0 | 0 | — |
case-16 | fail→pass | 13,518 | 6,887 | -49% | 1 | 1 | 0% | 2,015 | 1,287 | -36% | 0 | 0 | — |
case-17 | pass→pass | 12,729 | 8,609 | -32% | 1 | 1 | 0% | 1,794 | 1,703 | -5% | 0 | 0 | — |
case-18 | pass→pass | 10,742 | 7,144 | -33% | 1 | 1 | 0% | 1,805 | 1,386 | -23% | 0 | 0 | — |
case-19 | pass→pass | 6,889 | 4,144 | -40% | 1 | 1 | 0% | 1,119 | 884 | -21% | 0 | 0 | — |
case-20 | pass→pass | 14,179 | 11,973 | -16% | 1 | 1 | 0% | 3,115 | 3,174 | +2% | 0 | 0 | — |
case-21 | pass→pass | 7,555 | 6,790 | -10% | 1 | 1 | 0% | 1,332 | 1,456 | +9% | 0 | 0 | — |
case-22 | pass→pass | 11,791 | 10,485 | -11% | 1 | 1 | 0% | 2,135 | 2,142 | +0% | 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.
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.