adversarial-prompting
Warn
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION] (MEDIUM): The skill specifies the execution of a local file at 'scripts/export_analysis.py' to export results. As the source code for this script is missing, its safety regarding file system access or potential side effects cannot be verified.
- [PROMPT_INJECTION] (LOW): The skill is susceptible to Indirect Prompt Injection (Category 8) due to the lack of input isolation and the presence of execution capabilities.
- Ingestion points: The 'user's problem' is ingested via Phase 1 in SKILL.md.
- Boundary markers: Absent. No delimiters or isolation instructions are defined in SKILL.md to separate user-provided content.
- Capability inventory: The skill calls a Python script (
scripts/export_analysis.py) as described in SKILL.md. - Sanitization: Absent. No methods to sanitize or escape user input before processing are described.
Audit Metadata