optimize-png

Pass

Audited by Gen Agent Trust Hub on May 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides instructions to execute local binaries for image processing.\n
  • Evidence: SKILL.md contains bash command templates for oxipng and pngquant.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it processes file and directory paths provided by the user and incorporates them into shell commands.\n
  • Ingestion points: The Input parameter in SKILL.md accepts user-supplied files or directories.\n
  • Boundary markers: The commands use double quotes around variables (e.g., "") to prevent basic shell splitting, but no explicit instructions exist to disregard instructions embedded in the file metadata or to escape complex shell metacharacters.\n
  • Capability inventory: The skill has the capability to run arbitrary shell commands if the input paths are maliciously crafted to escape the provided quoting.\n
  • Sanitization: No specific sanitization or validation of the input paths is performed by the skill's instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
May 2, 2026, 07:54 PM
Security Audit — agent-trust-hub — optimize-png