Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use for everyday coding tasks that involve writing or modifying source code.
.claude/skills/galaxy-dawn-daily-coding/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-12 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 146% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 111% | 0% |
| case-10 | ✓→✗ | ▼ Worse | -50% | 0% |
| case-02 | ✓→✗ | ▼ Worse | -56% | 0% |
A minimal coding quality assurance checklist ensuring every code modification follows best practices.
Use this skill for:
Do not use this skill for:
bug-detective, architecture-design, or verification-loopany in TypeScriptpython# ❌ Don't def process(data=[]): # Mutable default argument pass # ✅ Should def process(data: list | None = None): data = data or []
python# ❌ Don't except: # Bare except pass # ✅ Should except ValueError as e: logger.error(f"Processing failed: {e}") raise
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | pass→fail | 8,788 | 5,977 | -32% | 1 | 1 | 0% | 1,443 | 727 | -50% | 0 | 0 | — |
case-01 | fail→fail | 12,967 | 4,069 | -69% | 1 | 1 | 0% | 2,867 | 649 | -77% | 0 | 0 | — |
case-02 | pass→fail | 7,182 | 4,399 | -39% | 1 | 1 | 0% | 1,528 | 670 | -56% | 0 | 0 | — |
case-03 | fail→fail | 15,369 | 15,367 | -0% | 1 | 1 | 0% | 2,535 | 3,431 | +35% | 0 | 0 | — |
case-04 | pass→pass | 12,352 | 16,009 | +30% | 1 | 1 | 0% | 2,423 | 2,940 | +21% | 0 | 0 | — |
case-05 | pass→fail | 2,288 | 4,421 | +93% | 1 | 1 | 0% | 385 | 628 | +63% | 0 | 0 | — |
case-06 | pass→fail | 9,409 | 3,831 | -59% | 1 | 1 | 0% | 1,800 | 664 | -63% | 0 | 0 | — |
case-07 | pass→fail | 4,976 | 4,688 | -6% | 1 | 1 | 0% | 938 | 680 | -28% | 0 | 0 | — |
case-08 | pass→fail | 10,244 | 4,122 | -60% | 1 | 1 | 0% | 2,205 | 725 | -67% | 0 | 0 | — |
case-09 | pass→pass | 4,764 | 7,414 | +56% | 1 | 1 | 0% | 805 | 1,870 | +132% | 0 | 0 | — |
case-11 | pass→pass | 14,910 | 21,075 | +41% | 1 | 1 | 0% | 3,206 | 5,513 | +72% | 0 | 0 | — |
case-12 | fail→pass | 11,024 | 7,716 | -30% | 1 | 1 | 0% | 1,708 | 1,685 | -1% | 0 | 0 | — |
case-13 | pass→fail | 9,135 | 5,672 | -38% | 1 | 1 | 0% | 1,589 | 780 | -51% | 0 | 0 | — |
case-14 | pass→pass | 8,091 | 9,022 | +12% | 1 | 1 | 0% | 1,473 | 2,080 | +41% | 0 | 0 | — |
case-15 | fail→fail | 5,796 | 3,728 | -36% | 1 | 1 | 0% | 1,092 | 661 | -39% | 0 | 0 | — |
case-16 | fail→fail | 4,066 | 4,097 | +1% | 1 | 1 | 0% | 232 | 631 | +172% | 0 | 0 | — |
case-17 | pass→fail | 5,913 | 4,538 | -23% | 1 | 1 | 0% | 1,175 | 658 | -44% | 0 | 0 | — |
case-18 | pass→pass | 7,779 | 9,906 | +27% | 1 | 1 | 0% | 1,445 | 2,341 | +62% | 0 | 0 | — |
case-19 | pass→fail | 14,923 | 5,062 | -66% | 1 | 1 | 0% | 3,304 | 720 | -78% | 0 | 0 | — |
case-20 | fail→pass | 3,482 | 6,963 | +100% | 1 | 1 | 0% | 684 | 1,686 | +146% | 0 | 0 | — |
case-21 | fail→pass | 6,984 | 15,322 | +119% | 1 | 1 | 0% | 1,264 | 2,661 | +111% | 0 | 0 | — |
case-22 | fail→fail | 7,034 | 5,912 | -16% | 1 | 1 | 0% | 1,304 | 1,651 | +27% | 0 | 0 | — |
case-23 | pass→fail | 5,243 | 4,810 | -8% | 1 | 1 | 0% | 488 | 677 | +39% | 0 | 0 | — |
case-24 | fail→fail | 7,366 | 3,592 | -51% | 1 | 1 | 0% | 1,507 | 624 | -59% | 0 | 0 | — |
case-25 | pass→fail | 6,726 | 4,084 | -39% | 1 | 1 | 0% | 1,420 | 726 | -49% | 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, and 10 counted toward the lift figure. The other 15 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 -32 percentage points is the difference between those two pass rates over the 10 comparable cases. 12 cases got worse with the skill loaded, and they are 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.