The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions provide workflows involving the execution of local CLI tools for ad platform management (e.g., node tools/clis/google-ads.js) and media processing utilities like ffmpeg.
[EXTERNAL_DOWNLOADS]: The documentation references downloading third-party tools and repositories from GitHub (e.g., jamiepine/voicebox) and utilizing npx to fetch and execute packages like remotion and create-video.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to ingest and process external performance data from CSV files, manual pastes, or API outputs to drive its creative generation logic.
Ingestion points: Performance data provided via CSV, text pastes, or external API tool outputs (documented in SKILL.md).
Boundary markers: No specific delimiters or "ignore embedded instructions" directives are defined for the ingested performance data.
Capability inventory: The skill can generate ad copy, suggest CLI commands for ad platform management (google-ads, meta-ads, etc.), and propose React-based video templates for rendering via Remotion.
Sanitization: There is no evidence of sanitization or validation of the input data before it is processed by the agent to generate new creative variations.

ad-creative