add-photo-certificate-watermarks
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill processes all files locally using the provided Python script, ensuring that sensitive documents are not uploaded to external services.
- [DATA_EXPOSURE]: The skill includes strong privacy protections, such as automatically removing EXIF metadata (which can contain GPS location and device data) and explicitly instructing the agent not to include sensitive details like ID numbers in its responses.
- [EXTERNAL_DOWNLOADS]: The skill relies on reputable, well-known libraries (
PillowandPyMuPDF) which are installed from standard package registries. - [COMMAND_EXECUTION]: Shell commands are used strictly for environment checks and executing the local watermarking script with specific, limited arguments.
- [INDIRECT_PROMPT_INJECTION]: While the skill ingests external files (images and PDFs), the risk is minimized because the workflow focuses on visual rendering rather than semantic interpretation of the document content, and includes a mandatory manual preview step.
Audit Metadata