ai-report-assist

Pass

Audited by Gen Agent Trust Hub on Oct 5, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process clinical histories and audio recordings, which are untrusted external inputs. These sources could be used for indirect prompt injection to manipulate reporting outcomes or agent behavior. \n
  • Ingestion points: Patient clinical data in get_structured_report and audio files in transcribe_dictation (SKILL.md). \n
  • Boundary markers: None identified; the skill does not specify delimiters to isolate patient data from system instructions. \n
  • Capability inventory: The skill uses the requests library for network operations and the open() function for file system access. \n
  • Sanitization: No sanitization or validation logic is provided for the external inputs. \n- [DATA_EXFILTRATION]: The skill defines workflows for sending sensitive medical information, including patient IDs and clinical histories, to external service domains such as api.radai.ai, api.abba.ai, and api.deeprad.ai. \n- [COMMAND_EXECUTION]: The skill provides Python code snippets that execute network requests and file reads, defining the functional capabilities accessible to the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 5, 2026, 05:08 PM