Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when documenting a clinical encounter as a progress note: organize it into the four SOAP sections in fixed order with each piece of data placed in the correct section — which cheaper models do not do by default.
.claude/skills/soap-progress-note-format/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 99% | 75 |
| gemini-3.1-pro-previewlowest | 80% | 5 |
| Model | Lift | Δ tokens | Δ turns | Cases | Verified |
|---|---|---|---|---|---|
| gemini-3.6-flashbest | +67% | +35% | 0% | 22 | 54d ago |
| gemini-3.5-flash | pending re-run | — | |||
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | — | — |
| case-03 | ✗→✓ | ▲ Improved | — | — |
| case-10 | ✗→✓ | ▲ Improved | — | — |
| case-11 | ✗→✓ | ▲ Improved | — | — |
| case-19 | ✗→✓ | ▲ Improved | — | — |
Enforce the SOAP progress-note format for any clinical progress/follow-up note: the note MUST be divided into exactly four labeled sections in the fixed order Subjective, Objective, Assessment, Plan, and each piece of information MUST land in its correct section. Apply whenever asked to write, document, or write up an encounter, visit, or progress note for the chart / medical record.
Subjective, Objective, Assessment,Plan (in that order). The single-letter abbreviations S: O: A: P: are acceptable as headers. Never write the note as one continuous narrative paragraph, and never substitute a different header scheme such as History / Exam / Diagnosis / Treatment / Recommendations.
here: chief complaint, history of present illness, symptom descriptions in the patient's terms, pertinent review of systems, relevant history, and adherence. Nothing you measured or observed.
numeric results of labs / imaging / tests go here — and ONLY here. Report the raw values; do not state the diagnosis in this section.
differential, and the clinical reasoning (severity, stability, progress) go here. Not raw data, and not the treatment orders.
frequency where given), referrals, patient education, disposition, and follow-up timing go here — not the diagnosis itself.
Base default (narrative — wrong) → SOAP (right):
> BEFORE: "Mr. C returns for his COPD. He says he is coughing more and gets winded walking to the > mailbox. His oxygen saturation is 92% and he has scattered wheezes. This looks like a COPD flare, so > I'll start prednisone and an antibiotic and see him back in two weeks."
> AFTER: > Subjective: Increased cough; dyspnea on minimal exertion (winded walking to the mailbox). > Objective: SpO2 92% on room air; scattered expiratory wheezes bilaterally. > Assessment: Acute COPD exacerbation. > Plan: Start oral prednisone taper and azithromycin; return in 2 weeks; return sooner if worsening.
Misallocation (wrong) → corrected: placing measured data in the wrong section.
> BEFORE (wrong): Subjective: "Blood pressure 150/94, knee swollen and tender on exam." > AFTER (right): those are measured/observed, so they belong in Objective: "BP 150/94; right knee with > effusion and joint-line tenderness." The Subjective section instead carries what the patient said > ("aching right knee, worse with stairs").
Assessment; the Plan may be one combined block or aper-problem plan — but keep the single four-section shell.
150/95") stays in Subjective because it is reported; a value you record in clinic goes in Objective.
Objective may hold only what you could verify, or note "examdeferred"; do not migrate reported symptoms into Objective to fill the section.
Objective (e.g., "Exam unremarkable"), not omitted andnot merged into another section.
Objective. Don't put them under Subjective.Assessment. Don't bury it in Objective.Plan. Don't list them inAssessment.
Subjective. Don't recast them as Objectivefindings.
Subjective.Objective section.Assessment instead of Plan.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
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 +67 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
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 | 7/10/2026 | +96% |
Other measured skills in the registry, with their headline benchmark lift.