food-label-review
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the AI agent to execute a bundled Node.js script (
scripts/generate_report.js) via shell commands to convert audit results from JSON into Word documents. This requires the agent to execute code in the host environment and perform local file read/write operations. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input in the form of food label text, images, and documents. This creates a vulnerability surface where maliciously crafted input could attempt to influence the agent's audit findings or the content of the final report.
- Ingestion points: User-provided food label materials described in the audit stages of
SKILL.md. - Boundary markers: The instructions do not define clear delimiters or warnings to separate user-provided data from standard compliance logic.
- Capability inventory: The agent can write local JSON files and execute a Node.js script that interacts with the file system.
- Sanitization: No explicit sanitization, validation, or escaping procedures for the ingested food label data are specified.
Audit Metadata