Portable Agent Skills for building with OpenMed — the on-device, Apache-2.0 clinical & biomedical NLP library. Each skill is a folder with a SKILL.md that works unchanged in Claude Code , OpenAI Codex , OpenCode , and compatible agents. Drop them in and your coding agent learns to wire up OpenMed pipelines — de-identification, NER, FHIR export, evaluation — plus the upstream/downstream healthcare tasks around them.
72 skills across 14 categories.
Get running in one command The same SKILL.md folders work unchanged in Claude Code , OpenAI Codex , OpenCode , and compatible clients of the open standard . The installer uses each client's documented skills directory plus the cross-client ~/.agents/skills convention:
git clone https://github.com/maziyarpanahi/openmed && cd openmed
./install-skills.sh # -> Claude Code, Codex, OpenCode, and ~/.agents/skills Or target one agent:
Agent
Command
Skills directory
Claude Code ./install-skills.sh claude~/.claude/skills/
OpenAI Codex ./install-skills.sh codex~/.codex/skills/
OpenCode ./install-skills.sh opencode~/.config/opencode/skills/
Cross-client convention ./install-skills.sh agents~/.agents/skills/
No clone? Copy the folders directly with cp -r skills/*/ ~/.claude/skills/ (swap the path per agent). Claude Code users can also install as a plugin, no clone needed:
/plugin marketplace add maziyarpanahi/openmed
/plugin install openmed-skills@openmed-skills
After installing, just ask your agent in plain language — it finds the right skill and writes correct, on-device OpenMed code for you:
You: Build a local OpenMed pipeline that de-identifies a synthetic discharge note and extracts medication entities. Keep the example synthetic.
Your agent loads deidentifying-clinical-text + extracting-clinical-entities and produces:
import openmed
note = "Synthetic patient Jane Example (MRN 12345), seen 2024-03-02, started on metformin 500mg BID."
deid = openmed .deidentify (
note , method = "mask" , policy = "hipaa_safe_harbor"
)
meds = openmed .analyze_text (
deid .deidentified_text ,
model_name = "pharma_detection_superclinical" ,
)→ The sample is synthetic. After the one-time model download, inference runs locally. Keep real PHI out of cloud-agent prompts, logs, and copied examples.
New here? Start with building-with-openmed
Legend: → before runs upstream of OpenMed, after → consumes its output, ↔ adjacent is a neighbouring task.
OpenMed core — build with OpenMed directly
Skill
What it does
Pairs
building-with-openmedOrient and bootstrap any project that uses OpenMed, the on-device clinical and biomedical NLP library, for named-entity recognition, PHI ….
↔ adjacent
choosing-openmed-modelsDiscover and pick the right OpenMed model for a clinical or biomedical task, domain, or language.
↔ adjacent
deidentifying-clinical-textRemove, mask, or replace PHI/PII in clinical free text on-device with OpenMed's deidentify().
↔ adjacent
extracting-clinical-entitiesRun clinical and biomedical named-entity recognition on medical text with OpenMed's analyze_text.
↔ adjacent
extracting-pii-entitiesDetect PHI/PII spans in clinical text with OpenMed's extract_pii without altering the text.
↔ adjacent
loading-openmed-modelsLoad OpenMed clinical/biomedical NER models from the Hugging Face Hub or a local path and reuse them efficiently across calls.
↔ adjacent
reidentifying-textReversibly de-identify clinical text with OpenMed and later restore the original PHI from a saved mapping.
↔ adjacent
running-zeroshot-nerExtract arbitrary, custom entity types from clinical or biomedical text with no fine-tuning using OpenMed's GLiNER / GLiNER2 zero-shot su….
↔ adjacent
Data ingestion — feed clinical text into OpenMed
Skill
What it does
Pairs
extracting-dicom-metadataReads DICOM file headers and DICOM-SR (Structured Report) content to pull study/series metadata and embedded report text, and flags PHI c….
→ before
fetching-fhir-resourcesFetches and pages FHIR R4 resources (Patient, DocumentReference, DiagnosticReport, Observation, Condition) from a FHIR REST server, decod….
→ before
generating-synthea-dataGenerates synthetic but realistic patient records (FHIR R4 bundles, C-CDA documents, CSV) with MITRE Synthea for development, CI fixtures….
↔ adjacent
parsing-ccda-documentsParses C-CDA / CCD XML clinical documents to extract human-readable section narrative plus coded entries, keyed by section LOINC codes an….
→ before
parsing-hl7v2-messagesDecodes pipe-delimited HL7 v2.x messages (ADT, ORU, MDM, ORM) into structured segments/fields/components and surfaces OBX-5 and NTE-3 fre….
→ before
De-identification & privacy
Skill
What it does
Pairs
auditing-deid-leakageAdversarially scan already-de-identified clinical text for residual identifiers and emit a leakage report that blocks release on any hit.
after →
auditing-deidentification-runsProduce a signed, reproducible, no-PHI audit trail for an OpenMed de-identification run via deidentify(audit=True).
after →
auditing-safe-harbor-checklistVerify OpenMed de-identified output against all 18 HIPAA Safe Harbor identifier categories and report residual re-identification risk.
after →
configuring-privacy-policiesSelect and customize OpenMed's seven bundled privacy policy profiles for de-identification, and build custom surrogate generators.
↔ adjacent
deidentifying-multilingual-textDe-identify non-English clinical text on-device with OpenMed by passing lang= and locale= to deidentify().
↔ adjacent
generating-synthetic-surrogatesReplace detected PHI with realistic, type-matched fake values in OpenMed so clinical notes stay readable and parseable instead of full of….
after →
pseudonymizing-for-gdprApply GDPR-grade pseudonymization to clinical or personal text with OpenMed, keeping a separately-held re-linkage key so the data can be ….
after →
reviewing-reidentification-riskRun expert-determination-style quasi-identifier risk scoring (k-anonymity, l-diversity) plus OpenMed's empirical re-identification attack….
after →
shifting-clinical-datesApply consistent per-patient date shifting in OpenMed that preserves intervals between events while satisfying HIPAA Safe Harbor's date rule.
after →
Clinical NLP — refine OpenMed output
Skill
What it does
Pairs
extracting-sdohExtracts social determinants of health (SDOH) — housing instability, food insecurity, unemployment, transportation barriers, social isola….
after →
parsing-lab-valuesParse laboratory values and reference ranges from clinical text and flag results as low, normal, high, or critical with OpenMed.
after →
reconciling-problem-listsDeduplicate and reconcile OpenMed-extracted conditions into one clean active problem list with clinical status (active / resolved / histo….
after →
resolving-clinical-contextAssign negation, temporality, and uncertainty (the ConText axes) to clinical entities extracted by OpenMed, so "denies chest pain" is n….
after →
segmenting-clinical-sectionsSplit a clinical note into canonical sections (Chief Complaint, HPI, PMH, Medications, Allergies, Assessment & Plan, etc.) before running….
→ before
summarizing-clinical-notesProduces structured, citation-anchored summaries of clinical notes — one-liner, hospital course, and problem-oriented views — where every….
after →
Skill
What it does
Pairs
coding-hcc-risk-adjustmentMaps chronic conditions extracted by OpenMed to CMS-HCC V28 risk-adjustment categories and estimates a RAF (Risk Adjustment Factor) score….
after →
coding-icd10Suggests candidate ICD-10-CM diagnosis codes (and ICD-10-PCS procedure codes) for diagnoses and procedures extracted by OpenMed, with rat….
after →
linking-umls-conceptsLinks entities extracted by OpenMed to UMLS Metathesaurus CUIs using the USER'S OWN UTS API key, with nothing from the Metathesaurus bund….
after →
mapping-loincMaps laboratory and clinical observation names extracted by OpenMed to LOINC codes using the public Regenstrief LOINC and FHIR terminolog….
after →
mapping-to-snomedMaps clinical concept spans extracted by OpenMed to SNOMED CT concepts through a USER-SUPPLIED terminology server (the user's own Ontoser….
after →
normalizing-rxnormNormalizes drug mentions extracted by OpenMed to RxNorm RxCUIs using the free public RxNav/RxNorm REST API.
after →
Skill
What it does
Pairs
assembling-fhir-bundlesPackage multiple FHIR R4 resources produced from OpenMed output into a single valid transaction Bundle ready to POST to an EHR, using Ope….
after →
bridging-presidio-and-spacyCombine OpenMed clinical NLP with Microsoft Presidio, spaCy, or LangChain through OpenMed's built-in interop adapter registry (openmed.in….
↔ adjacent
exporting-bulk-fhirKick off and harvest a FHIR Bulk Data $export (system-, group-, or patient-level) and stream the resulting NDJSON into a batch OpenMed de….
→ before
exporting-to-fhirConvert OpenMed NER output (entities from openmed.analyze_text) into FHIR R4 resources — Condition, MedicationStatement, Observation — us….
after →
querying-terminology-serviceCall a user-supplied FHIR terminology server ($validate-code, $expand, $lookup, $translate) to validate and expand clinical codes without….
↔ adjacent
scaffolding-smart-on-fhirScaffold a SMART-on-FHIR app (SMART App Launch v2 — EHR launch and standalone launch, OAuth2 PKCE, scopes, token handling, fhirContext) s….
↔ adjacent
validating-us-coreValidate FHIR R4 resources and Bundles against US Core / USCDI profiles with the official HL7 FHIR validator before submitting to an EHR.
after →
Skill
What it does
Pairs
auditing-subgroup-fairnessAudit an OpenMed NER or de-identification model for performance disparities across demographic subgroups (sex, age band, race/ethnicity w….
↔ adjacent
authoring-model-cardsGenerate a model card for an OpenMed clinical NER or de-identification model documenting intended use, quantitative metrics, subgroup per….
after →
benchmarking-clinical-nerScore an OpenMed clinical or biomedical NER model against a user-supplied gold corpus with entity-level precision, recall, and F1, then b….
↔ adjacent
building-gold-corpusScaffold a synthetic gold-standard annotation project for evaluating OpenMed NER and de-identification models — label schema, annotation ….
↔ adjacent
evaluating-with-leakage-gatesEvaluate an OpenMed de-identification or clinical NER model against the leakage-first release gates G1a through G8, which gate releases o….
↔ adjacent
gating-deid-leakageAdd a CI gate that fails the build when an OpenMed de-identification model's recall on a held-out PHI set drops below threshold or any cr….
↔ adjacent
Skill
What it does
Pairs
annotating-variantsAnnotates VCF variants and normalizes HGVS nomenclature with public, license-free annotators (Ensembl VEP REST, VEP/SnpEff/ANNOVAR offlin….
↔ adjacent
defining-cohort-phenotypesAuthors computable phenotype and cohort definitions in the OHDSI ATLAS / CIRCE style over the OMOP CDM, combining standard concept sets w….
↔ adjacent
mining-pubmed-literatureSearches and fetches PubMed and PMC via NCBI E-utilities (ESearch then EFetch/ESummary) to gather biomedical evidence and build text corpora.
↔ adjacent
parsing-trial-eligibilityParses free-text clinical-trial eligibility criteria into structured inclusion and exclusion logic, then matches them against patient fac….
after →
searching-clinicaltrialsSearches ClinicalTrials.gov for studies by condition, intervention, and recruitment status using the modern v2 REST API with cursor (page….
↔ adjacent
Skill
What it does
Pairs
extracting-lab-tablesDetects and extracts tabular laboratory panels from PDFs, scans, and images into structured rows ready for OpenMed and FHIR.
→ before
ingesting-clinical-documentsTurn scanned faxes, images, and CSV/CDA exports into clean text ready for OpenMed de-identification and NER, fully on-device.
→ before
structuring-radiology-reportsConverts free-text radiology narratives into structured findings and impression — with measurements, laterality, anatomy, and follow-up r….
after →
Skill
What it does
Pairs
auditing-part11-trailsGenerates and verifies 21 CFR Part 11-style audit trails — who/what/when, electronic signatures, and tamper-evidence — for OpenMed pipeli….
↔ adjacent
checking-hipaa-complianceRuns a HIPAA Privacy and Security Rule checklist over a data pipeline and produces a gap report before deploying OpenMed on PHI.
↔ adjacent
Safety & pharmacovigilance
Skill
What it does
Pairs
detecting-pv-signalsComputes disproportionality signals — PRR, ROR, EBGM, and IC (BCPNN) — over FAERS / OpenFDA drug-event data to flag potential safety signals.
↔ adjacent
querying-openfda-labelsLooks up FDA drug labels, NDC directory entries, indications, boxed warnings, and recalls/enforcement actions via the free public OpenFDA….
↔ adjacent
reporting-adverse-eventsStructures adverse-event mentions that OpenMed extracts into FAERS / ICH E2B(R3) reportable fields — suspect drug, reaction (MedDRA PT), ….
after →
Skill
What it does
Pairs
building-patient-timelinesAssemble a chronological patient timeline from OpenMed-extracted clinical events, normalizing dates and resolving relative time expressio….
after →
computing-ecqmsCompute electronic clinical quality measures (eCQMs) over structured data using CQL/QDM logic, lifting note-derived numerator and exclusi….
after →
etl-to-omop-cdmMap OpenMed-extracted, terminology-coded conditions, drugs, and measurements into OMOP CDM v5.4 clinical tables (condition_occurrence, dr….
after →
Skill
What it does
Pairs
batch-processing-clinical-textRun large-scale batch NER, PII extraction, or de-identification over many clinical notes on-device with OpenMed, with sharding, checkpoin….
↔ adjacent
deploying-openmed-mcpRun OpenMed's Model Context Protocol (MCP) server so coding agents (Claude Code, Codex) and chat clients can call clinical NER, PII extra….
↔ adjacent
enforcing-nophi-loggingAdd a logging and telemetry guard that scrubs or blocks PHI from logs, traces, and error reports around an OpenMed deployment.
↔ adjacent
running-openmed-ondeviceRun OpenMed models fully on-device with the MLX (Apple Silicon), CoreML (iOS/macOS), or ONNX/WebGPU (cross-platform/browser) backends, in….
↔ adjacent
serving-openmed-rest-apiStand up OpenMed's FastAPI REST service for clinical NER, PII extraction, and de-identification, with health checks, model keep-alive/unl….
↔ adjacent
Skill
What it does
Pairs
benchmark-pii-recallBenchmark an OpenMed PII model with synthetic gold spans and report label-aware exact-span and grapheme recall without emitting identifie….
deidentify-a-datasetDe-identify selected free-text columns in a local CSV, JSONL, or Parquet dataset with OpenMed and produce a separate redacted dataset plu….
extract-clinical-entities-to-fhirExtract clinical entities from synthetic or already de-identified text with OpenMed and map them into deterministic FHIR R4 resources and….
pick-a-pii-modelSelect an on-device OpenMed PII model from the committed registry by language, runtime format, and size budget, then require recall valid….
Every skill follows the open spec: kebab-case folder name matching name, a description that states what it does and when to use it , a body under 500 lines, and detail pushed into references/. Validate and regenerate this catalog with:
python skills/build_catalog.py --check # CI gate# rewrite README + marketplace