Skill UI

This guide provides a comprehensive framework for authors to articulate the unique and marginal contribution of their research within the highly specialized field of health economics. It teaches how to frame findings against canonical literature, identify specific gaps, and structure the argument to satisfy expert reviewers, ensuring the paper's novelty is clearly demonstrated.

This skill helps diagnose the health and gather comprehensive status reports for the Ruflo system, including the MCP server and active agents. It provides a read-only workflow to check system status or, when explicitly requested, performs a non-destructive repair and reports the changes.

This tool provides a comprehensive, multi-faceted audit of a codebase's health. It scores the project across critical areas like security vulnerabilities, exposed secrets, dependency risks, and structural integrity. Ideal for senior engineers performing pre-commit or pre-release quality assessments to identify weak spots and generate a prioritized remediation plan.

This skill provides a comprehensive checklist for auditing de-identified Protected Health Information (PHI) against the 18 categories required by HIPAA Safe Harbor (45 CFR 164.514(b)(2)). It maps detected PII entities to these categories, identifies gaps (e.g., age > 89, rare geography), and guides the assessment of residual re-identification risk. This is essential for proving the completeness and compliance of de-identification efforts.

This skill acts as a comprehensive map for developing clinical and biomedical NLP pipelines using OpenMed. OpenMed is a local-first Python library that performs named-entity recognition (NER), protected health information (PHI) de-identification, and FHIR export, all running fully on-device without cloud calls. Use this guide when starting a new clinical text project, needing to select the appropriate capability (e.g., zero-shot NER, auditing, or FHIR conversion), or requiring an entry point for the core OpenMed features.

This skill guides the creation of computable phenotypes and patient cohorts using the OHDSI/OMOP CDM standards. It allows users to define precise, executable criteria by combining standard concept sets (like PheKB) with advanced Natural Language Processing (NLP) features extracted from free clinical text. Use this when structured codes alone are insufficient to capture critical health information (e.g., symptoms, social context).

A powerful, on-device tool for removing, masking, or replacing sensitive Protected Health Information (PHI/PII) from clinical free text. It ensures patient privacy by running fully locally, supporting methods like masking, hashing, and date shifting. Essential for preparing medical records for safe sharing, storage, or analysis, adhering to standards like HIPAA and GDPR.

This skill de-identifies sensitive PII from clinical notes across numerous languages, including Spanish, German, French, and Arabic. It automatically selects language-specific PII models, national ID regex patterns, and generates locale-aware fake surrogates. Designed for global healthcare datasets, it runs entirely on-device, ensuring maximum privacy and linguistic accuracy for diverse international clinical records.

This skill implements a comprehensive logging guard to prevent Protected Health Information (PHI) leakage into logs, traces, and error reports. It acts as a pre-emission filter, redacting structured identifiers (SSN, email, phone) via regex and handling free-text PHI using advanced models. Essential for maintaining strict data privacy and compliance when shipping logs to centralized stores or error trackers.

This skill evaluates de-identification and clinical NER models against rigorous leakage-first release gates (G1a-G8). Instead of relying on aggregate F1 scores, it focuses on detecting residual Protected Health Information (PHI) leakage and enforcing per-label recall floors, which is crucial for HIPAA compliance. Determines if a model is RELEASABLE or QUARANTINED.

This skill facilitates the extraction of population-scale (cohort-level) clinical text from Electronic Health Records (EHR) or FHIR data warehouses using the FHIR Bulk Data Access ($export) operation. It manages the asynchronous job kickoff, polling, and downloading of NDJSON files. The resulting data stream is designed to be fed in batches into OpenMed for mandatory de-identification and Named Entity Recognition (NER) analysis.

Use openmed.extract_pii to detect and retrieve specific types of Protected Health Information (PHI) or Personally Identifiable Information (PII) spans (names, dates, MRNs, etc.) from clinical text. Unlike deidentification, this function returns the original text and the detected spans, allowing users to audit, preview redactions, normalize labels, or route spans to a custom anonymizer without altering the source data. It is ideal for pre-processing and policy enforcement.