seo-images
Fail
Audited by Gen Agent Trust Hub on Jul 27, 2026
Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the user to run 'sudo apt install' for system dependencies, which constitutes a privilege escalation risk. It also executes numerous shell commands such as 'exiftool', 'cwebp', 'convert', and 'ffmpeg' using arguments derived from user-provided file paths and metadata, creating a risk for command injection.
- [REMOTE_CODE_EXECUTION]: The skill references and executes local Python scripts ('scripts/parse_html.py', 'iptc_ai_label.py') via an external tool 'claude-seo'. Since these files are not present in the skill package, their behavior is unverifiable and could execute arbitrary code.
- [EXTERNAL_DOWNLOADS]: The skill recommends downloading and installing software from external repositories and refers to an external GitHub repository for licensing and documentation.
- [PROMPT_INJECTION]: The instructions contain trigger patterns that guide agent behavior based on specific user inputs, which could be manipulated to bypass intended constraints if not properly delimited.
Recommendations
- AI detected serious security threats
Audit Metadata