alterlab-clinical-reports — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited alterlab-clinical-reports (Agent Skill) and scored it 92/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 2 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 2 flagged
A bulleted imperative like {match} tells the agent to never reveal, disclose, or mention something to the user. Used adversarially it can instruct the agent to hide its tool calls or lie about what it did — stripping the transparency a user relies on to trust the agent.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Document medical information with precision, accuracy, and regulatory compliance. This skill covers four report families: case reports for journal publication, diagnostic reports for clinical practice (radiology, pathology, lab), clinical trial reports for regulatory submission (SAE, CSR), and patient documentation for medical records (SOAP, H&P, discharge).
Critical principle: clinical reports must be accurate, complete, objective, and compliant with applicable regulations (HIPAA, FDA, ICH-GCP). Patient privacy and data integrity are paramount.
Use when:
assets/ for that report type.scripts/ validator before sign-out.Follow the CARE (CAse REport) checklist: title, keywords, structured abstract, introduction, patient information, clinical findings, timeline, diagnostic assessment, therapeutic interventions, follow-up/outcomes, discussion, patient perspective, and informed consent. Mind journal-specific limits (word count, figures, reference style) and de-identify before submission. → Element-by-element detail, journal requirements, and the 18 HIPAA identifiers: references/care_report_sections.md. High-level checklist: references/case_report_guidelines.md.
Each uses a standardized section structure (demographics → indication → technique → comparison → findings → impression for radiology; gross/microscopic/diagnosis for pathology; results with reference ranges and critical-value reporting for lab). Use structured-reporting templates (BI-RADS, Lung-RADS, CAP synoptic) where they apply. → Full section templates: references/diagnostic_report_templates.md. Standards and lexicons (ACR, CAP, LOINC): references/diagnostic_reports_standards.md.
SAE reports document serious adverse events with causality and expectedness, on strict regulatory timelines (7/15 days). CSRs follow the ICH-E3 section structure for regulatory submission. Protocol deviations are categorized (minor/major/violation) with CAPA documentation. → Component-by-component structures: references/clinical_trial_report_structures.md. Regulatory framing (ICH-E3, CONSORT, timelines): references/clinical_trial_reporting.md.
SOAP notes for progress, H&P for admission/initial encounters, discharge summaries for handoff to outpatient providers. Use standard abbreviations, sign and date, document medical necessity for billing. → Format structures: references/patient_record_formats.md. Coding and documentation guidance: references/patient_documentation.md.
identifiers) or Expert Determination; Business Associate Agreements for third parties.
Part 312 (IND).
audit trails, investigator responsibilities.
⚠️ Caveat — automated de-identification is NOT a compliance guarantee. The bundled scripts/check_deidentification.py is a pure regex scan. Pattern matching has known, substantial false-negative rates: it misses unconventional name spellings, free-text dates, narrative addresses, rare identifiers, and anything outside its fixed patterns. It is a rough first-pass screen only — not a substitute for line-by-line manual review by a qualified person, and not a validated de-identification tool (e.g., Microsoft Presidio, Philter, or a certified Expert Determination). Passing this script does not establish HIPAA Safe Harbor compliance and must never be relied upon as a privacy guarantee. Always perform manual review before any disclosure or publication.Detailed guidance: references/regulatory_compliance.md.
Use standardized nomenclature: SNOMED CT (clinical terms), LOINC (lab/clinical observations), ICD-10-CM (diagnosis coding), CPT (procedure coding). Respect the Joint Commission "Do Not Use" abbreviation list (e.g. write "unit" not "U", always use a leading zero, never a trailing zero). Comprehensive standards: references/medical_terminology.md.
Tables for demographics, adverse events, lab values over time, and efficacy outcomes; figures for Kaplan-Meier curves, forest plots, CONSORT flow diagrams, and case-report timelines. Images must be ≥300 dpi, de-identified, with consent for recognizable patients. Detail: references/data_presentation.md.
Documentation must be complete, accurate, timely, clear, and compliant. Use the per-type validation checklists (CARE, diagnostic completeness, SAE regulatory compliance, billing requirements) and the scripts/ validators.
review → journal selection/submission → revision.
→ peer review (complex cases) → sign-out → critical-value notification. STAT <1h, routine 24-48h.
submit to sponsor/IRB/FDA → follow-up to resolution (24h-15 days).
review → QC → approval/submission (6-12 months post-completion).
references/)care_report_sections.md — CARE element-by-element, journal requirements, 18 HIPAA identifierscase_report_guidelines.md — CARE guidelines, journal requirements, writing tipsdiagnostic_report_templates.md — radiology/pathology/lab section templatesdiagnostic_reports_standards.md — ACR, CAP, laboratory reporting standardsclinical_trial_report_structures.md — SAE/CSR/deviation component structuresclinical_trial_reporting.md — ICH-E3, CONSORT, SAE reporting, CSR structurepatient_record_formats.md — SOAP/H&P/discharge section formatspatient_documentation.md — SOAP, H&P, discharge, codingregulatory_compliance.md — HIPAA, 21 CFR Part 11, ICH-GCP, FDAmedical_terminology.md — SNOMED, LOINC, ICD-10, abbreviationsdata_presentation.md — tables, figures, safety data, CONSORT diagramspeer_review_standards.md — review criteria for clinical manuscriptsassets/)case_report_template.md, radiology_report_template.md, pathology_report_template.md, lab_report_template.md, clinical_trial_sae_template.md, clinical_trial_csr_template.md, soap_note_template.md, history_physical_template.md, discharge_summary_template.md, consult_note_template.md, quality_checklist.md, hipaa_compliance_checklist.md.
scripts/)validate_case_report.py, validate_trial_report.py, check_deidentification.py, format_adverse_events.py, generate_report_template.py, extract_clinical_data.py, compliance_checker.py, terminology_validator.py.
Pairs with scientific-writing (clear medical prose), peer-review (quality assessment), citation-mgmt (literature references), research-grants (protocol development), and literature-review (background sections).
literature review, overgeneralization from a single case.
correlation, delayed critical-value notification.
unreported deviations, selective reporting.
missing medical necessity, unsigned/undated notes.
Before finalizing any clinical report, verify:
Final note: clinical report quality directly impacts patient safety, healthcare delivery, and medical knowledge. Always prioritize accuracy, privacy, and professionalism.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.