image
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides practical examples for using common shell utilities, including
cwebp,mogrify,jpegoptim, andcurl, to perform image optimization and metadata extraction. These commands are typical for developer and marketing workflows. - [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to fetch content from external websites via
curlto identify image tags. This establishes an ingestion point for potentially untrusted data that could contain malicious instructions. 1. Ingestion points: The standard output of thecurlcommand when analyzing external URLs inSKILL.md. 2. Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded content in the captured output. 3. Capability inventory: The skill utilizes network access throughcurland local execution of image processing binaries. 4. Sanitization: Filtering is performed viagrep -oPpatterns, which reduces the volume of data but does not provide robust protection against sophisticated prompt injection attacks embedded in the source HTML.
Audit Metadata