The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external performance data and local marketing context files to generate and iterate on ad creative.
Ingestion points: CSV or API-sourced performance metrics in Mode 2 and context from .agents/product-marketing.md.
Boundary markers: The instructions do not define specific delimiters or directives to ignore instructions within the ingested data.
Capability inventory: The skill uses local CLI tools (google-ads.js) and network operations via curl for API interactions.
Sanitization: There are no explicit steps for sanitizing or validating external data before it is interpolated into prompts.\n- [REMOTE_CODE_EXECUTION]: The skill provides shell command examples for the user to execute, such as curl for interacting with external APIs (including Gemini and ElevenLabs) and node for executing local ad platform management scripts.\n- [EXTERNAL_DOWNLOADS]: The documentation references external tools and repositories, including a GitHub repository for the Voicebox voice synthesis studio.\n- [DYNAMIC_EXECUTION]: The skill utilizes Remotion, a framework that renders video by executing React code at runtime. It encourages the use of data-driven templates where external JSON data is injected into React components for batch production.

ad-creative