Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build and analyze phylogenetic trees using MAFFT (multiple alignment), IQ-TREE 2 (maximum likelihood), and FastTree (fast NJ/ML). Visualize with ETE3 or FigTree. For evolutionary analysis, microbial genomics, viral phylodynamics, protein family analysis, and molecular clock studies.
.claude/skills/mkurman-phylogenetics/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 3% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 1% | 0% |
-|-------------|---------| | GTR+G4 | General Time Reversible + Gamma | Most flexible DNA model | | HKY+G4 | Hasegawa-Kishino-Yano + Gamma | Two-rate model (common) | | TrN+G4 | Tamura-Nei | Unequal transitions | | JC | Jukes-Cantor | Simplest; all rates equal |
| Model | Description | Use case | |-------|-------------|---------| | LG+G4 | Le-Gascuel + Gamma | Best average protein model | | WAG+G4 | Whelan-Goldman | Widely used | | JTT+G4 | Jones-Taylor-Thornton | Classical model | | Q.pfam+G4 | pfam-trained | For Pfam-like protein families | | Q.bird+G4 | Bird-specific | Vertebrate proteins |
Tip: Use -m TEST to let IQ-TREE automatically select the best model.
linsi for small (<200 seq), fftns or auto for large alignments-m TEST for IQ-TREE unless you have a specific reason-B 1000) for branch supportRDP4, GARD)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,989 | 10,931 | -27% | 1 | 1 | 0% | 2,586 | 2,728 | +5% | 0 | 0 | — |
case-02 | fail→fail | 16,972 | 13,564 | -20% | 1 | 1 | 0% | 2,683 | 2,877 | +7% | 0 | 0 | — |
case-03 | fail→fail | 16,225 | 11,127 | -31% | 1 | 1 | 0% | 2,573 | 2,387 | -7% | 0 | 0 | — |
case-04 | pass→pass | 14,729 | 10,763 | -27% | 1 | 1 | 0% | 2,433 | 2,512 | +3% | 0 | 0 | — |
case-05 | fail→fail | 11,363 | 5,873 | -48% | 1 | 1 | 0% | 1,896 | 1,568 | -17% | 0 | 0 | — |
case-06 | pass→pass | 8,991 | 5,466 | -39% | 1 | 1 | 0% | 1,486 | 1,508 | +1% | 0 | 0 | — |
case-07 | pass→pass | 12,203 | 11,964 | -2% | 1 | 1 | 0% | 1,901 | 2,427 | +28% | 0 | 0 | — |
case-08 | pass→pass | 15,194 | 11,832 | -22% | 1 | 1 | 0% | 2,405 | 2,597 | +8% | 0 | 0 | — |
case-09 | fail→pass | 6,872 | 4,878 | -29% | 1 | 1 | 0% | 1,160 | 1,473 | +27% | 0 | 0 | — |
case-10 | fail→fail | 8,851 | 6,922 | -22% | 1 | 1 | 0% | 1,558 | 1,848 | +19% | 0 | 0 | — |
case-11 | fail→fail | 6,327 | 6,637 | +5% | 1 | 1 | 0% | 1,176 | 1,911 | +63% | 0 | 0 | — |
case-12 | pass→pass | 5,020 | 4,507 | -10% | 1 | 1 | 0% | 816 | 1,402 | +72% | 0 | 0 | — |
case-13 | fail→fail | 13,045 | 2,998 | -77% | 1 | 1 | 0% | 2,051 | 1,102 | -46% | 0 | 0 | — |
case-14 | fail→pass | 23,401 | 5,044 | -78% | 1 | 1 | 0% | 2,019 | 1,385 | -31% | 0 | 0 | — |
case-15 | pass→pass | 10,129 | 4,618 | -54% | 1 | 1 | 0% | 1,640 | 1,371 | -16% | 0 | 0 | — |
case-16 | fail→pass | 14,826 | 4,400 | -70% | 1 | 1 | 0% | 905 | 1,395 | +54% | 0 | 0 | — |
case-17 | pass→pass | 7,221 | 2,747 | -62% | 1 | 1 | 0% | 1,214 | 1,067 | -12% | 0 | 0 | — |
case-22 | pass→pass | 21,977 | 24,202 | +10% | 1 | 1 | 0% | 3,455 | 4,576 | +32% | 0 | 0 | — |
case-18 | pass→pass | 12,390 | 7,772 | -37% | 1 | 1 | 0% | 2,002 | 1,908 | -5% | 0 | 0 | — |
case-19 | pass→pass | 13,721 | 10,543 | -23% | 1 | 1 | 0% | 2,163 | 2,323 | +7% | 0 | 0 | — |
case-20 | pass→pass | 10,855 | 3,902 | -64% | 1 | 1 | 0% | 1,733 | 1,220 | -30% | 0 | 0 | — |
case-21 | pass→pass | 12,935 | 16,268 | +26% | 1 | 1 | 0% | 2,259 | 3,597 | +59% | 0 | 0 | — |
case-23 | pass→pass | 16,206 | 15,253 | -6% | 1 | 1 | 0% | 2,837 | 3,354 | +18% | 0 | 0 | — |
case-24 | pass→pass | 14,603 | 15,684 | +7% | 1 | 1 | 0% | 2,680 | 3,480 | +30% | 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. 24 cases were attempted, and 23 counted toward the lift figure. The other 1 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 +13 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.