The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The media_proxy.py and critique_scripts.py scripts facilitate communication with external AI and media generation APIs, including FAL and ElevenLabs. These requests are routed through GooseWorks-managed proxy endpoints (e.g., api.gooseworks-ai.com domains) to handle authentication and credit management. The skill also utilizes urllib.request.urlretrieve to download generated assets locally.
[COMMAND_EXECUTION]: The skill uses subprocess.run in verify_handoff.py and several test files to execute internal Python components like lint_scripts.py and prepare_angle_context.py. This is part of the skill's self-validation and linting workflow to ensure generated scripts meet specific quality and technical criteria.
[DATA_EXFILTRATION]: The media_proxy.py script reads authentication data from the configuration file ~/.gooseworks/credentials.json. This information, along with campaign context and ad script content, is transmitted to vendor-authorized proxy endpoints for the purpose of generating media and critiques. This behavior is consistent with the skill's primary function and relies on the developer's own infrastructure.
[INDIRECT_PROMPT_INJECTION]: The critique_scripts.py script processes ad candidates, customer quotes, and creative briefs to generate prompts for an external LLM critic. While this creates a surface for indirect prompt injection via the ingested content, the skill uses structured JSON formatting and explicit role definitions for the LLM to maintain process integrity.

write-video-ad-script