generate-image
Warn
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: MEDIUMPERSISTENCEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PERSISTENCE]: The skill instructions direct the agent to store sensitive API keys in the user's shell configuration files (e.g., ~/.bashrc, ~/.zshrc). This is intended to maintain session credentials but involves modifying persistent system environment settings.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user prompts and inserts them into JSON request bodies for external API services, creating an indirect injection surface.\n
- Ingestion points: User-provided descriptions via the argument-hint and workflow in SKILL.md.\n
- Boundary markers: No explicit markers or delimiters are defined in the API request templates.\n
- Capability inventory: The skill performs network requests to external APIs, writes image files to the local file system, and modifies shell configuration files (SKILL.md).\n
- Sanitization: Guidelines advise the agent to escape special characters for JSON safety, providing basic sanitization.\n- [EXTERNAL_DOWNLOADS]: The skill retrieves data and images from OpenAI and Google Gemini APIs. As these are well-known technology services and trusted organizations, these references are documented as part of the core functionality.\n- [COMMAND_EXECUTION]: Automating the persistence of credentials to shell profiles requires the execution of shell commands to append export statements to system files.
Audit Metadata