Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Medical code mapping and classification tools. ICD-10-CM/PCS, CPT, SNOMED CT, HCPCS, LOINC, RxNorm. Code validation, mapping between terminologies, HCC risk adjustment, and reimbursement modeling.
.claude/skills/mkurman-medical-coding/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -36% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | -20% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -18% | 0% |
Medical code mapping and classification: ICD-10-CM/PCS, CPT, SNOMED CT, HCPCS, LOINC, RxNorm. Covers code validation, cross-terminology mapping, HCC risk adjustment, and reimbursement modeling used in healthcare billing and clinical research.
pythonhcc_map = { "E11.9": "HCC 19", # Diabetes without complications "I10": "HCC 134", # Essential hypertension "N18.3": "HCC 138", # CKD stage 3 } def calc_hcc(codes): hccs = set() for c in codes: if c in hcc_map: hccs.add(hcc_map[c]) return list(hccs) print(calc_hcc(["E11.9", "I10"]))
pythonvalid_icd10 = set() # from official CMS file or lookup def validate_code(code): if code in valid_icd10: return True, "Valid" if code[:3] in valid_icd10: return True, "Valid (category)" return False, "Unknown code"
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,210 | 6,898 | +11% | 1 | 1 | 0% | 1,167 | 1,555 | +33% | 0 | 0 | — |
case-02 | fail→fail | 7,704 | 8,335 | +8% | 1 | 1 | 0% | 1,479 | 1,972 | +33% | 0 | 0 | — |
case-03 | pass→pass | 5,204 | 2,137 | -59% | 1 | 1 | 0% | 924 | 738 | -20% | 0 | 0 | — |
case-04 | pass→pass | 6,483 | 2,491 | -62% | 1 | 1 | 0% | 1,063 | 875 | -18% | 0 | 0 | — |
case-05 | fail→pass | 9,220 | 3,310 | -64% | 1 | 1 | 0% | 1,563 | 993 | -36% | 0 | 0 | — |
case-06 | fail→pass | 9,017 | 3,142 | -65% | 1 | 1 | 0% | 1,513 | 927 | -39% | 0 | 0 | — |
case-07 | fail→fail | 11,635 | 9,882 | -15% | 1 | 1 | 0% | 2,056 | 2,076 | +1% | 0 | 0 | — |
case-08 | pass→pass | 6,449 | 4,628 | -28% | 1 | 1 | 0% | 900 | 1,204 | +34% | 0 | 0 | — |
case-09 | fail→fail | 6,199 | 7,310 | +18% | 1 | 1 | 0% | 1,080 | 1,597 | +48% | 0 | 0 | — |
case-10 | pass→pass | 7,663 | 5,500 | -28% | 1 | 1 | 0% | 1,379 | 1,375 | -0% | 0 | 0 | — |
case-11 | pass→pass | 11,048 | 8,684 | -21% | 1 | 1 | 0% | 1,952 | 1,890 | -3% | 0 | 0 | — |
case-12 | pass→pass | 7,008 | 6,170 | -12% | 1 | 1 | 0% | 1,135 | 1,531 | +35% | 0 | 0 | — |
case-13 | fail→pass | 10,330 | 5,146 | -50% | 1 | 1 | 0% | 1,686 | 1,131 | -33% | 0 | 0 | — |
case-14 | pass→pass | 11,352 | 11,996 | +6% | 1 | 1 | 0% | 1,942 | 2,431 | +25% | 0 | 0 | — |
case-15 | pass→pass | 5,143 | 5,373 | +4% | 1 | 1 | 0% | 813 | 1,177 | +45% | 0 | 0 | — |
case-16 | pass→pass | 4,350 | 4,022 | -8% | 1 | 1 | 0% | 551 | 951 | +73% | 0 | 0 | — |
case-17 | pass→pass | 13,469 | 10,891 | -19% | 1 | 1 | 0% | 2,233 | 2,225 | -0% | 0 | 0 | — |
case-18 | pass→pass | 8,308 | 7,802 | -6% | 1 | 1 | 0% | 1,331 | 1,842 | +38% | 0 | 0 | — |
case-19 | pass→pass | 9,311 | 8,364 | -10% | 1 | 1 | 0% | 1,762 | 2,044 | +16% | 0 | 0 | — |
case-20 | pass→pass | 3,082 | 4,854 | +57% | 1 | 1 | 0% | 467 | 1,073 | +130% | 0 | 0 | — |
case-21 | pass→pass | 5,030 | 3,557 | -29% | 1 | 1 | 0% | 832 | 1,041 | +25% | 0 | 0 | — |
case-22 | pass→pass | 3,287 | 3,041 | -7% | 1 | 1 | 0% | 496 | 871 | +76% | 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. 22 cases were attempted. The headline lift of +14 percentage points is the difference between those two pass rates over the 22 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.