Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Validates all code changes before committing by running format, lint, type, and unit test checks. Use after making backend (Python) or frontend (TypeScript) changes, before committing or finishing a work session.
.claude/skills/streamlit-checking-changes/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -56% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -63% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -63% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -77% | 0% |
Run at the end of a work session or after completing a set of changes — not after every small edit.
bashmake check
This runs formatting, linting, type checking, Knip unused-export and unused-dependency analysis, and unit tests on all uncommitted files (staged, unstaged, and untracked).
make checkmake checkmake check succeedsE2E_CHECK=true make check to also run changed e2e tests| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 21,075 | 130,698 | +520% | 1 | 1 | 0% | 1,569 | 691 | -56% | 0 | 0 | — |
case-02 | fail→pass | 15,900 | 5,535 | -65% | 1 | 1 | 0% | 2,391 | 851 | -64% | 0 | 0 | — |
case-03 | fail→pass | 16,592 | 10,039 | -39% | 1 | 1 | 0% | 2,521 | 943 | -63% | 0 | 0 | — |
case-04 | pass→pass | 11,195 | 4,149 | -63% | 1 | 1 | 0% | 1,612 | 615 | -62% | 0 | 0 | — |
case-05 | pass→pass | 35,831 | 7,509 | -79% | 1 | 1 | 0% | 1,926 | 526 | -73% | 0 | 0 | — |
case-06 | pass→pass | 9,870 | 2,174 | -78% | 1 | 1 | 0% | 916 | 413 | -55% | 0 | 0 | — |
case-07 | pass→pass | 11,419 | 3,208 | -72% | 1 | 1 | 0% | 1,329 | 619 | -53% | 0 | 0 | — |
case-08 | pass→pass | 76,965 | 4,242 | -94% | 1 | 1 | 0% | 2,101 | 604 | -71% | 0 | 0 | — |
case-09 | pass→pass | 28,474 | 5,256 | -82% | 1 | 1 | 0% | 1,631 | 507 | -69% | 0 | 0 | — |
case-10 | pass→pass | 26,334 | 14,044 | -47% | 1 | 1 | 0% | 2,473 | 1,796 | -27% | 0 | 0 | — |
case-11 | fail→pass | 21,262 | 5,081 | -76% | 1 | 1 | 0% | 2,214 | 829 | -63% | 0 | 0 | — |
case-12 | fail→pass | 77,209 | 2,733 | -96% | 1 | 1 | 0% | 2,360 | 544 | -77% | 0 | 0 | — |
case-13 | pass→pass | 9,751 | 2,657 | -73% | 1 | 1 | 0% | 1,635 | 492 | -70% | 0 | 0 | — |
case-14 | pass→pass | 82,566 | 4,400 | -95% | 1 | 1 | 0% | 1,843 | 788 | -57% | 0 | 0 | — |
case-15 | pass→pass | 9,078 | 2,999 | -67% | 1 | 1 | 0% | 1,322 | 552 | -58% | 0 | 0 | — |
case-16 | pass→pass | 9,239 | 2,832 | -69% | 1 | 1 | 0% | 1,251 | 493 | -61% | 0 | 0 | — |
case-17 | pass→pass | 9,192 | 3,805 | -59% | 1 | 1 | 0% | 1,227 | 525 | -57% | 0 | 0 | — |
case-18 | fail→pass | 14,619 | 3,526 | -76% | 1 | 1 | 0% | 902 | 557 | -38% | 0 | 0 | — |
case-19 | pass→pass | 9,001 | 2,999 | -67% | 1 | 1 | 0% | 1,318 | 484 | -63% | 0 | 0 | — |
case-20 | pass→pass | 6,379 | 6,276 | -2% | 1 | 1 | 0% | 1,032 | 859 | -17% | 0 | 0 | — |
case-21 | pass→pass | 15,155 | 10,650 | -30% | 1 | 1 | 0% | 2,448 | 2,149 | -12% | 0 | 0 | — |
case-22 | pass→pass | 14,889 | 4,326 | -71% | 1 | 1 | 0% | 878 | 743 | -15% | 0 | 0 | — |
case-23 | pass→pass | 8,829 | 7,020 | -20% | 1 | 1 | 0% | 1,269 | 1,170 | -8% | 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. 23 cases were attempted. The headline lift of +26 percentage points is the difference between those two pass rates over the 23 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.