seo-images
Warn
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: MEDIUMPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PRIVILEGE_ESCALATION]: The skill documentation includes
sudocommand recommendations within the error handling section (e.g.,sudo apt install libimage-exiftool-perl). While intended as user instructions for dependency management, the presence ofsudowithin an agent's skill instructions presents a risk if an autonomous agent attempts to execute the commands directly to resolve missing dependencies. - [COMMAND_EXECUTION]: The skill relies on several shell-based utilities for image processing and metadata management, specifically
exiftool,cwebp,convert(ImageMagick), andffmpeg. These tools are invoked to perform format conversions, resizing, and metadata injection. - [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection.
- Ingestion points: Data is ingested from external web pages during image audits and from JSON files within the
.seo-cache/directory (e.g.,site-meta.json,page-analysis.json). - Boundary markers: The instructions lack explicit boundary markers or instructions to treat external data as untrusted when processing it for analysis.
- Capability inventory: The skill can execute shell commands (
exiftool,cwebp,convert,ffmpeg) and write results to the local filesystem (.seo-cache/). - Sanitization: The instructions do not specify sanitization or escaping for data extracted from web pages (like alt text, titles, or descriptions) before it is passed as arguments to shell commands for metadata injection (e.g.,
exiftool -IPTC:ObjectName="[extracted_text]").
Audit Metadata