Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when reporting a pharmacogenetic result — a drug-metabolizing-enzyme phenotype or an allele's functional status: emit the CPIC/PharmGKB consensus term vocabulary (e.g. 'normal metabolizer' not 'extensive metabolizer', 'no function' not 'null') — which cheaper models do not do by default.
.claude/skills/cpic-pharmacogenetic-terms/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 3 |
| Model | Lift | Δ tokens | Δ turns | Cases | Verified |
|---|---|---|---|---|---|
| gemini-3.6-flashbest | +5% | +67% | 0% | 22 | 54d ago |
| gemini-3.5-flash | pending re-run | — | |||
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | — | — |
| case-03 | ✓→✓ | = Same ✓ | — | — |
| case-13 | ✓→✓ | = Same ✓ | — | — |
| case-01 | ✓→✓ | = Same ✓ | — | — |
| case-08 | ✗→✗ | = Same ✗ | — | — |
Report every pharmacogenetic result using ONLY the CPIC/PharmGKB consensus term vocabulary, spelled exactly. This applies whenever you translate a genotype, diplotype, or activity score into a phenotype call, or describe what an allele does. Do not invent, abbreviate, or fall back to legacy laboratory terms.
R1 — Metabolizer phenotype is exactly one of five terms. For a drug-metabolizing enzyme (CYP2D6, CYP2C19, CYP2C9, DPYD, TPMT, ...), the phenotype call is one of, and only one of:
The full-activity middle category is normal metabolizer. The historical term "extensive metabolizer" is retired and must never appear. "rapid metabolizer" is a distinct category sitting between ultrarapid and normal; it is not a synonym for ultrarapid.
R2 — Allele functional status is exactly one of six terms. An allele's function is one of:
A complete loss-of-function allele is no function (never "null", "null allele", "nonfunctional", or "loss-of-function"). A partially reduced allele is decreased function (never "reduced function" or "partial function"). A gain-of-activity allele is increased function (never "gain-of-function"). The reference/fully-active allele is normal function (never "wild-type" or "fully functional"). Use unknown function when the allele is uncharacterized and uncertain function when the evidence conflicts — these are two distinct terms.
R3 — CYP2D6 activity score maps to phenotype at fixed band boundaries. Sum the per-allele activity values, then map:
An activity score of exactly 1.0 is an intermediate metabolizer, not normal — this is the current consensus (a score of 1.0 was moved out of the normal band). The CYP2D610 allele contributes an activity value of 0.25 (a decreased-function allele), not 0.5.
R4 — Never substitute legacy or colloquial synonyms. "extensive metabolizer", "efficient/deficient metabolizer", "fast/slow metabolizer", "wild-type", "null allele", "loss-of-function" (as a status label), "reduced function", "gain-of-function", "hyperactive", "nonfunctional", and "inactive" are all non-conforming. Map each back to the R1/R2 terms.
R5 — Other result types (for completeness).
poor function — note the transporter bucket is "poor function", not "poor metabolizer".
metabolizer phenotype and never "carrier" or "at-risk".
BEFORE: "extensive metabolizer" / "EM" -> AFTER: "normal metabolizer"
BEFORE: "null metabolizer" / "non-metabolizer" -> AFTER: "poor metabolizer"
BEFORE: "normal metabolizer" -> AFTER: "intermediate metabolizer" (1.0 falls in the intermediate band)
BEFORE: "rapid metabolizer" -> AFTER: "ultrarapid metabolizer"
BEFORE: "null" / "loss-of-function allele" -> AFTER: "no function"
BEFORE: "gain-of-function" -> AFTER: "increased function"
BEFORE: "reduced function" -> AFTER: "decreased function"
BEFORE: "wild-type" / "fully functional" -> AFTER: "normal function"
ultrarapid metabolizer. Do not collapse them into one.
score of exactly 2.0.
studies is uncertain function. Pick per the evidence — they are not interchangeable.
enzymes only.
retired.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
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 +5 percentage points is the difference between those two pass rates over the 22 comparable cases.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.5-flash | verified | 7/9/2026 | +4% |
Other measured skills in the registry, with their headline benchmark lift.