skills/anthropics/healthcare/fhir/Gen Agent Trust Hub

fhir

Pass

Audited by Gen Agent Trust Hub on Jul 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • Command Execution for Data Processing: The skill uses shell commands like bun to execute local extraction scripts and rm to clean up temporary directories. These are used strictly for processing document attachments and maintaining system hygiene, with explicit instructions to use only system-provided paths to prevent directory traversal.
  • External Dependency Management: Instructions mention installing dependencies for the doc-extract utility during the first use. This is a standard setup procedure for the skill's bundled processing tools.
  • Clinical Data Ingestion Controls: The skill identifies clinical note text as untrusted data. It implements a defense-in-depth strategy by delegating the analysis of this text to a separate, tool-restricted environment (clinical-note-extract) to prevent any embedded instructions from influencing agent behavior.
  • Sensitive Data Handling: While the skill interacts with HIPAA-protected health information, it operates as a local orchestrator and does not exfiltrate data to external services. It provides clear guidance on secure configuration management, such as using environment variables in .mcp.json for persistent connections.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 7, 2026, 08:37 AM
Security Audit — agent-trust-hub — fhir