The Agent Skills Directory

[SAFE]: The skill is instructional in nature and does not include any scripts, external dependencies, or commands that interact with the file system or network. It guides the agent in using approved internal tools for image generation purposes.\n- [PROMPT_INJECTION]: The skill processes user-provided inputs to create image prompts. While this is an indirect injection surface, the skill includes specific instructions to enforce safety guidelines, such as ensuring all actors are adults and specifying modest clothing to prevent inappropriate content generation. Ingestion points: User-provided briefs for actor descriptions. Boundary markers: Explicit instructions for the agent to use specific fictionality statements and default to safety-compliant descriptions. Capability inventory: Limited to the image_generate and job_status tools. Sanitization: Guidance for the agent to avoid words that trigger content filters and to describe clothing coverage positively to avoid model bypass.

generating-ai-actors