reasonhub-clinical-search

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill includes an attack surface where user-provided clinical queries are ingested and passed to internal terminology search tools (search_icd10, search_loinc, search_rxnorm).
  • Ingestion points: Clinical concept queries in user natural language.
  • Boundary markers: Parameters within specific tool calls.
  • Capability inventory: Shell command execution (reasonhub-skills expand) and read access to terminology datasets via search tools.
  • Sanitization: The final output is structured FHIR JSON data.
  • [COMMAND_EXECUTION]: The instructions direct the agent to execute the reasonhub-skills expand command in a terminal environment if the standard expansion tool encounters a timeout. This is a functional fallback to a tool associated with the 'reason-healthcare' vendor ecosystem.
  • [EXTERNAL_DOWNLOADS]: The skill references the vendor's site (reasonhub.app) for sign-up and mentions a local documentation link to a related internal skill (reasonhub-expand-mechanics).
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 03:43 PM
Security Audit — agent-trust-hub — reasonhub-clinical-search