Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →You are a technical debt expert specializing in identifying, quantifying, and prioritizing technical debt in software projects. Analyze the codebase to uncover debt, assess its impact, and create acti
.claude/skills/dokhacgiakhoa-codebase-cleanup-tech-debt/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-25 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 4% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 84% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 81% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 16% | 0% |
You are a technical debt expert specializing in identifying, quantifying, and prioritizing technical debt in software projects. Analyze the codebase to uncover debt, assess its impact, and create actionable remediation plans.
The user needs a comprehensive technical debt analysis to understand what's slowing down development, increasing bugs, and creating maintenance challenges. Focus on practical, measurable improvements with clear ROI.
$ARGUMENTS
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 29,386 | 41,650 | +42% | 1 | 1 | 0% | 5,152 | 5,342 | +4% | 0 | 0 | — |
case-02 | pass→pass | 10,466 | 17,593 | +68% | 1 | 1 | 0% | 2,060 | 3,782 | +84% | 0 | 0 | — |
case-03 | pass→pass | 12,894 | 22,547 | +75% | 1 | 1 | 0% | 2,942 | 5,335 | +81% | 0 | 0 | — |
case-04 | pass→pass | 14,899 | 14,517 | -3% | 1 | 1 | 0% | 2,443 | 2,824 | +16% | 0 | 0 | — |
case-05 | pass→pass | 13,961 | 14,460 | +4% | 1 | 1 | 0% | 2,205 | 2,881 | +31% | 0 | 0 | — |
case-06 | pass→pass | 10,289 | 13,621 | +32% | 1 | 1 | 0% | 1,679 | 2,606 | +55% | 0 | 0 | — |
case-07 | pass→pass | 11,836 | 14,530 | +23% | 1 | 1 | 0% | 1,757 | 2,379 | +35% | 0 | 0 | — |
case-08 | fail→fail | 17,247 | 12,066 | -30% | 1 | 1 | 0% | 2,436 | 2,284 | -6% | 0 | 0 | — |
case-09 | pass→pass | 12,655 | 12,121 | -4% | 1 | 1 | 0% | 1,739 | 2,019 | +16% | 0 | 0 | — |
case-10 | fail→fail | 8,899 | 9,236 | +4% | 1 | 1 | 0% | 1,511 | 1,916 | +27% | 0 | 0 | — |
case-11 | pass→pass | 9,801 | 9,417 | -4% | 1 | 1 | 0% | 1,695 | 1,927 | +14% | 0 | 0 | — |
case-12 | pass→pass | 42,704 | 11,956 | -72% | 1 | 1 | 0% | 2,026 | 2,174 | +7% | 0 | 0 | — |
case-13 | pass→pass | 13,256 | 19,364 | +46% | 1 | 1 | 0% | 2,135 | 2,960 | +39% | 0 | 0 | — |
case-14 | pass→pass | 9,488 | 9,241 | -3% | 1 | 1 | 0% | 1,433 | 1,949 | +36% | 0 | 0 | — |
case-15 | pass→pass | 20,670 | 17,369 | -16% | 1 | 1 | 0% | 3,149 | 3,140 | -0% | 0 | 0 | — |
case-16 | pass→pass | 12,527 | 13,866 | +11% | 1 | 1 | 0% | 2,041 | 2,725 | +34% | 0 | 0 | — |
case-17 | pass→pass | 15,662 | 18,705 | +19% | 1 | 1 | 0% | 2,376 | 2,945 | +24% | 0 | 0 | — |
case-18 | pass→pass | 7,280 | 8,938 | +23% | 1 | 1 | 0% | 1,273 | 1,816 | +43% | 0 | 0 | — |
case-19 | pass→pass | 10,672 | 9,802 | -8% | 1 | 1 | 0% | 1,445 | 1,869 | +29% | 0 | 0 | — |
case-20 | pass→pass | 17,181 | 21,461 | +25% | 1 | 1 | 0% | 2,727 | 3,431 | +26% | 0 | 0 | — |
case-21 | pass→pass | 14,470 | 15,412 | +7% | 1 | 1 | 0% | 1,940 | 2,480 | +28% | 0 | 0 | — |
case-22 | pass→pass | 14,192 | 17,172 | +21% | 1 | 1 | 0% | 2,408 | 2,918 | +21% | 0 | 0 | — |
case-23 | pass→pass | 15,633 | 19,719 | +26% | 1 | 1 | 0% | 2,356 | 3,280 | +39% | 0 | 0 | — |
case-24 | pass→pass | 14,084 | 13,450 | -5% | 1 | 1 | 0% | 2,096 | 2,349 | +12% | 0 | 0 | — |
case-25 | fail→pass | 14,029 | 12,875 | -8% | 1 | 1 | 0% | 2,111 | 2,366 | +12% | 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. 25 cases were attempted. The headline lift of +4 percentage points is the difference between those two pass rates over the 25 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.