The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a structured workflow that involves the execution of various local Python scripts to process media assets.
Evidence: SKILL.md instructs the agent to run scripts such as python scripts/image_generator.py, python scripts/audio_generator.py, and python scripts/video_compositor.py.
[PROMPT_INJECTION]: The multi-stage agent pipeline (11 distinct roles) processes user-provided product information to generate scripts and visual descriptions. This architecture is susceptible to indirect prompt injection if user inputs contain adversarial instructions designed to influence downstream agents.
Evidence: The '文案创作师智能体' in references/agent-prompts.md takes product category, selling points, and target audience as direct inputs, which then flow into subsequent stages like storyboarding and directing.
[EXTERNAL_DOWNLOADS]: The skill is designed to interface with external AI service providers for media generation.
Evidence: scripts/image_generator.py and scripts/audio_generator.py contain code placeholders (e.g., generate_single_image, download_image) for interacting with external APIs like Stable Diffusion or Azure Speech, intended to download generated assets during the workflow.

video-creation-collaborator