social-creative-designer

Pass

Audited by Gen Agent Trust Hub on Mar 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill requires the execution of system-level commands to function. It instructs the agent to perform a global package installation using pip3 install imageio[ffmpeg] --break-system-packages and to execute a generated Python script to compile images into MP4 video files. While these actions are standard for the skill's media-processing purpose, they involve direct shell and runtime interaction.\n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it takes arbitrary text from the user and interpolates it into complex prompts for an image generation tool.\n
  • Ingestion points: User inputs for 'post concept' and 'overlay text' collected during the intake phase in SKILL.md.\n
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent or the image generation tool to treat these user-supplied strings as non-executable data.\n
  • Capability inventory: The agent has access to the mcp__nanobanana__generate_image tool and a Python execution environment, which could be influenced by malicious instructions embedded in the user text.\n
  • Sanitization: The skill does not define any validation or sanitization steps to filter out potentially malicious content from the user inputs before prompt construction.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 31, 2026, 07:51 PM
Security Audit — agent-trust-hub — social-creative-designer