The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted external data from the 'inputs/' directory (customer reviews and ad comments) to drive its 'grounded generation' approach. This creates a surface for indirect prompt injection. 1. Ingestion points: Content is read from 'inputs/reviews/' and 'inputs/comments/' files. 2. Boundary markers: No specific delimiters or instructions are provided to the agent to treat this content as untrusted. 3. Capability inventory: The agent can manage live campaigns via tools for Google Ads, Meta Ads, and other platforms. 4. Sanitization: Input data is not sanitized, although output sanitization for the HTML preview tool is explicitly instructed.
[COMMAND_EXECUTION]: The skill documents the use of local CLI tools (e.g., 'node tools/clis/google-ads.js') for performance data retrieval and campaign management. This is a legitimate use of tool-calling capabilities.
[EXTERNAL_DOWNLOADS]: The skill's documentation references and provides examples for numerous well-known AI services and APIs (Google, OpenAI, ElevenLabs, etc.). These are trusted service providers and the references are for legitimate integration purposes.

ad-creative