img-gen-taste

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill instructions define a clear behavioral role for the agent as an art director. No patterns attempting to bypass safety filters or disregard prior instructions were found.
  • [DATA_EXFILTRATION]: No access to sensitive files or credentials was detected. The skill uses local metadata (tweet and image IDs) for style reference purposes only and performs no external network requests.
  • [REMOTE_CODE_EXECUTION]: The skill consists of Markdown and YAML configuration files. It does not download or execute external scripts or binaries.
  • [COMMAND_EXECUTION]: No shell commands or dynamic execution patterns (such as backtick execution) were identified in any of the skill files.
  • [INDIRECT_PROMPT_INJECTION]: While the skill processes user-supplied subject matter to generate prompts, it uses a structured YAML 'creative_spec' to maintain data integrity. The skill lacks dangerous capabilities (like file system writes or code execution) that could be exploited via malicious user input.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 08:27 AM
Security Audit — agent-trust-hub — img-gen-taste