photo-homework-a4
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from user-provided photos using an OCR tool (Read). The identified text is then interpolated into placeholders within an HTML template. This creates a surface for indirect prompt injection, as malicious instructions or scripts embedded in the handwritten text could be included in the generated file.
- Ingestion points: Homework photos processed by the Read tool (SKILL.md Step 1).
- Boundary markers: None explicitly defined in the template interpolation process.
- Capability inventory: The skill uses
present_filesto deliver the final HTML to the user; it does not exhibit network or sensitive file system access. - Sanitization: There are no instructions to sanitize or escape the OCR output before inserting it into the HTML structure.
- [DYNAMIC_EXECUTION]: The skill performs dynamic content generation by copying a local template (
assets/homework-a4-template.html) and performing string replacements to produce a final.htmlfile (SKILL.md Step 3). While the generated content is intended for display or printing, generating executable formats like HTML from untrusted input is a minor dynamic execution risk.
Audit Metadata