axiom-validate-screenshots
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
sipscommand-line utility within a Bash loop to retrieve image dimensions. This involves the execution of shell commands on file paths provided by the user and discovered via the Glob tool. - [DYNAMIC_EXECUTION]: The skill provides instructions for the agent to generate and run a Bash script at runtime to process batches of discovered screenshots for dimension verification.
- [INDIRECT_PROMPT_INJECTION]: The skill uses multimodal vision to analyze the text and content of user-supplied screenshots. This creates a surface for indirect prompt injection where malicious instructions embedded within images could attempt to manipulate the agent's logic or reporting.
- Ingestion points: User-specified directory containing screenshots (SKILL.md, Step 2/4).
- Boundary markers: No explicit delimiters or instructions to ignore image-embedded commands are provided to the vision model.
- Capability inventory: Shell command execution via
sips, file system globbing, and image reading via vision. - Sanitization: No sanitization of the visual content or extracted text is performed before it is processed by the model.
Audit Metadata