clinical-case-report

Pass

Audited by Gen Agent Trust Hub on May 8, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to parse untrusted user input (a clinical brief) and incorporate that data into a generated HTML file, which constitutes a surface for indirect prompt injection if malicious instructions or scripts are embedded in the input brief.\n
  • Ingestion points: Patient age, sex, chief complaint, and clinical narratives extracted from the user's prompt in Step 1.\n
  • Boundary markers: Absent; the workflow does not instruct the agent to use delimiters or explicit 'ignore instructions' markers when processing the user's brief.\n
  • Capability inventory: The skill possesses the capability to write a multi-section HTML document (index.html) using Step 6.\n
  • Sanitization: Instructions do not explicitly mandate HTML escaping or sanitization of patient-provided narrative strings before they are interpolated into the artifact.
Audit Metadata
Risk Level
SAFE
Analyzed
May 8, 2026, 02:54 PM