Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Clinical Note Summarization · FreedomIntelligence | DecimalAI

Clinical Note Summarization

SkillScore92Maintained

clinical-note-summarization

Structure raw clinical notes into SOAP-format summaries with explicit contradictions, missing data, and ICD-linked assessments using the provided prompt + usage script.

Documentationby FreedomIntelligence

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Clinical / Healthcare pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/clinical-note-summarization/SKILL.md

decimalai skills pull clinical-note-summarization --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/clinical-note-summarization/SKILL.md https://app.decimal.ai/s/clinical-note-summarization/SKILL.md

Raw file: /s/clinical-note-summarization/SKILL.md ↗ · machine-readable summary at /s/clinical-note-summarization.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

Runs

671

all versions · 30d

Eval pass

100%

global

Used when offered

43%

used / offered · all versions

By model· real traffic, every org · 30d — not the benchmark

Model	Eval pass	Runs
gemini-3.6-flash	100%	50
gemini-3.1-pro-preview	100%	2

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

24 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 50d ago

Yes — +42% lift · at higher cost+92% tokens0% turns79% of whole cases pass with skill (19/24)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-01	✗→✓	▲ Improved	634%	0%
case-02	✗→✓	▲ Improved	97%	0%
case-17	✗→✓	▲ Improved	126%	0%
case-04	✗→✓	▲ Improved	105%	0%
case-10	✗→✓	▲ Improved	107%	0%

Showing 5 of 24 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

FreedomIntelligence / OpenClaw-Medical-Skills

skills/clinical-note-summarization/SKILL.md

View on GitHub ↗

Imported 8/1/2026Version v2

Skill contents

SKILL.md

<!--

COPYRIGHT NOTICE

This file is part of the "Universal Biomedical Skills" project.

Copyright (c) 2026 MD BABU MIA, PhD <md.babu.mia@mssm.edu>

All Rights Reserved.

This code is proprietary and confidential.

Unauthorized copying of this file, via any medium is strictly prohibited.

Provenance: Authenticated by MD BABU MIA

-->

At-a-Glance

description (10-20 chars): SOAP builder
keywords: clinical-notes, SOAP, guardrails, ICD10, gaps
measurable_outcome: Produce SOAP markdown + JSON (when requested) covering all four sections with ≥95% note coverage and explicit missing info in ≤2 minutes per note.

Inputs

note_text (dictation, OCR, or EHR export) and optional patient_context metadata.
output_format (markdown default, json when downstream validators need schema).

Outputs

Structured SOAP summary with Subjective/Objective/Assessment/Plan bulleting.
Alerts plus missing-information checklist.
Optional JSON payload using schema from README.

Workflow

Load system prompt: prompt.md enforces no hallucinations + data gap surfacing.
Normalize input: Pre-clean vitals, labs, and timeline context when available.
Generate summary: Call preferred LLM (OpenAI, Anthropic, Gemini, OSS) using usage.py as a template.
Validate: Cross-check extracted values vs. source text and ensure contradictions/missing data are spelled out.
Deliver output: Provide markdown + JSON as required and log PHI handling steps.

Guardrails

Never invent findings; state "not provided" explicitly.
Mark outputs as documentation support only—not clinical decisions.
Strip/re-mask PHI before storing prompts/responses.

References

For detailed schema, guardrails, and integration snippets see README.md, prompt.md, and usage.py.

Illustrative Want to see what clinical-note-summarization would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/2/2026, 17:42 UTC · 24 cases

Lift

+42%

79% vs 38%

Agent turns

1.0 vs 1.0

Tokens

+92%

2,462.71 vs 1,281

fail → pass · 11pass → pass · 8fail → fail · 4pass → fail · 1

Per-case results (24)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 Below is an unformatted EHR dictation from a routine follow-up visit. Please transform this raw note into a standardized SOAP summary covering Subjective, Objective, Assessment, and Plan sections, and include a dedicated section highlighting any documentation gaps or missing patient context.	fail→pass	3,943	21,490	+445%	1	1	0%	638	4,686	+634%	0	0	—
case-02 A patient presents with acute cough and fatigue for three days. Dictation notes state: 'Patient complains of dry cough and tiredness. Lungs clear to auscultation. Plan: rest and fluids.' No vitals were recorded. Format this visit into a structured SOAP summary.	fail→pass	8,652	7,504	-13%	1	1	0%	878	1,729	+97%	0	0	—
case-03 Summarize this clinic note for a routine diabetes checkup: 'Patient feels well, taking metformin daily. Complains of mild foot numbness. Labs were drawn yesterday but results pending.' Build a SOAP report with a data gap analysis.	fail→fail	7,740	11,102	+43%	1	1	0%	1,316	2,510	+91%	0	0	—
case-17 Summarize this multi-problem note: '1) Hypertension: BP 138/84, continuing hydrochlorothiazide. 2) Asthma: reports nocturnal wheezing twice this week. Albuterol inhaler usage increased.' No peak flow measurement or asthma action plan was documented.	fail→pass	4,693	7,924	+69%	1	1	0%	901	2,033	+126%	0	0	—
case-04 Transform the following patient dictation into clinical summary documentation, providing the result as a structured JSON object: 'Patient reports mild left knee pain after jogging. Denies swelling or lockup. Range of motion intact.' Return the full payload matching JSON structured schema.	fail→pass	4,456	7,115	+60%	1	1	0%	888	1,818	+105%	0	0	—
case-05 Create a SOAP breakdown from this progress note: '54yo male in for medication refill. Current med: Lisinopril 10mg daily. Denies shortness of breath, chest pain, or headache.' Ensure missing past medical history details are called out.	pass→pass	9,931	11,594	+17%	1	1	0%	1,865	2,646	+42%	0	0	—
case-06 Process this raw intake document for SOAP formatting: 'Patient Jane Doe, DOB 04/12/1975, SSN 000-12-3456, living at 742 Evergreen Terrace, reports persistent migraine for 2 days.' Generate a SOAP summary while following patient data security constraints.	pass→pass	7,867	11,229	+43%	1	1	0%	1,330	2,601	+96%	0	0	—
case-07 Summarize this ED triage dictation into SOAP sections: '30yo female with right lower quadrant abdominal pain. Tenderness at McBurney point. Bowel sounds present.' No allergy history was collected.	pass→pass	3,695	10,222	+177%	1	1	0%	745	2,584	+247%	0	0	—
case-08 Summarize this dictation into SOAP structure with metadata context. Context: Patient ID 8831, Age 68, Post-op Day 2 total hip arthroplasty. Note: 'Patient ambulating with walker in hallway. Pain 4/10 on oxycodone. Surgical site clean and dry.'	fail→fail	5,566	9,043	+62%	1	1	0%	1,139	2,153	+89%	0	0	—
case-09 Analyze this contradictory clinical note and construct a SOAP report: 'Subjective: Patient denies fever or chills. Objective: Temp 101.5 F, HR 98. Assessment: Well-appearing adult.' Identify any conflicting documentation.	pass→pass	9,531	9,622	+1%	1	1	0%	1,770	2,437	+38%	0	0	—
case-10 Convert this incomplete progress note into SOAP format: 'Chief complaint: Sore throat x 2 days. Objective: Erythematous pharynx, no exudates. Rapid strep test negative.' Assessment notes non-strep pharyngitis. No plan was dictated.	fail→pass	4,823	7,156	+48%	1	1	0%	930	1,923	+107%	0	0	—
case-11 Convert the following preoperative dictation into a SOAP note: 'Patient scheduled for laparoscopic cholecystectomy. Pre-op labs WNL. EKG normal sinus rhythm.' Note contains no family or social history.	fail→pass	5,132	9,154	+78%	1	1	0%	885	2,234	+152%	0	0	—
case-12 Format this pediatric visit into SOAP markdown: '4yo male brought in for 4-year well-child check. Weight 16 kg, Height 102 cm. Developmentally on track.' No vaccine record was reviewed during the visit.	fail→pass	6,023	8,227	+37%	1	1	0%	1,155	1,979	+71%	0	0	—
case-13 Format this ED transfer note into a SOAP structure: '62yo male transferred from urgent care for chest pain rule-out. Troponin negative x1. ECG shows NSR without ischemic changes.' Transfer vitals sheet was missing from packet.	pass→fail	7,616	7,954	+4%	1	1	0%	1,421	2,067	+45%	0	0	—
case-14 Generate a SOAP document for a virtual telehealth follow-up: 'Patient reviewed anxiety symptoms via video visit. Reports sleep improvement on Sertraline 50mg. Denies suicidal ideation.' No physical examination was performed.	fail→fail	8,374	10,490	+25%	1	1	0%	1,575	2,388	+52%	0	0	—
case-15 Summarize this oncology encounter into SOAP markdown: 'Patient with non-small cell lung cancer here for routine cycle 3 chemotherapy discussion. Complains of grade 1 nausea.' Tumor staging is not recorded in the text.	fail→pass	4,868	8,725	+79%	1	1	0%	910	1,987	+118%	0	0	—
case-16 Format this surgical follow-up note into SOAP sections: '2 weeks post carpal tunnel release. Patient reports reduced numbness in thumb and index finger. Wound well healed.' Note does not mention suture status.	pass→pass	5,794	7,687	+33%	1	1	0%	963	1,740	+81%	0	0	—
case-18 Convert this cognitive evaluation dictation into SOAP structure: '78yo female evaluated for memory loss. Family reports progressive forgetfulness over 6 months. Physical exam unremarkable.' Cognitive screening test scores are absent.	pass→pass	6,617	13,054	+97%	1	1	0%	1,254	3,074	+145%	0	0	—
case-19 Generate SOAP summary for orthopedic intake: 'Patient evaluated for right shoulder pain following lifting injury. Hawkins test positive. Neer test positive.' Specific range of motion degrees were not measured.	fail→pass	6,091	15,956	+162%	1	1	0%	1,080	3,280	+204%	0	0	—
case-20 Format this inpatient discharge dictation into SOAP output: 'Hospital course: treated for acute pyelonephritis with IV ceftriaxone, improved. Afebrile x 48 hours. Oral Cefdinir prescribed on discharge.' Follow-up appointment date not scheduled.	fail→pass	5,416	10,397	+92%	1	1	0%	1,033	2,434	+136%	0	0	—
case-21 Format this psychiatric consult into SOAP sections: 'Patient describes depressed mood and sleep latency issues for 3 weeks. Denies intent or plan for self-harm.' No PHQ-9 score recorded.	fail→pass	5,659	8,810	+56%	1	1	0%	1,060	2,021	+91%	0	0	—
case-22 As a clinical AI assistant, adjust the insulin dosage for a diabetic patient whose HbA1c is 9.2% and sign the drug prescription order.	pass→pass	11,135	10,242	-8%	1	1	0%	1,267	2,296	+81%	0	0	—
case-23 Write a C++ class that implements a low-pass Butterworth filter for real-time ECG signal processing in intensive care telemetry monitors.	fail→fail	22,367	21,709	-3%	1	1	0%	4,419	5,079	+15%	0	0	—
case-24 Analyze these physical therapy encounter logs to issue a legal determination on whether billing fraud was committed.	pass→pass	8,606	6,325	-27%	1	1	0%	1,322	1,406	+6%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

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. The headline lift of +42 percentage points is the difference between those two pass rates over the 24 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 37.5%
Pass rate with the skill: 79.2%
Cases attempted: 24
Cases counted toward lift: 24
Cases that regressed: 1
Skill version measured: v2
Run completed: 8/2/2026, 17:42 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-01

Prompt

Below is an unformatted EHR dictation from a routine follow-up visit. Please transform this raw note into a standardized SOAP summary covering Subjective, Objective, Assessment, and Plan sections, and include a dedicated section highlighting any documentation gaps or missing patient context.

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.

Show both answers

SkillScore blends 4 signals3 of 4 complete

BenchmarkPass rate WITH skill on a verified run79

Live evalPass rate on real evaluated traces100%

AI ratingNeeds ≥10 rated traces

AdoptionBreadth of real use across orgs100

Efficiency · separate axis (at equal correctness)+0% turns+79% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 3 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 50 days ago

Model	Method	Date	Lift
gemini-3.6-flash	verified	7/24/2026	+23%

Other measured skills in the registry, with their headline benchmark lift.