The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests external performance data (CSV, API results, or pasted text) to iterate on creative variations, which creates a potential surface for indirect prompt injection.
Ingestion points: Performance data input sources are defined in the 'Iterating from Performance Data' section of SKILL.md.
Boundary markers: No specific delimiters are implemented to separate external data from system instructions.
Capability inventory: The skill possesses the capability to execute command-line tools for Google, Meta, and LinkedIn ads.
Sanitization: There is no evidence of input validation or sanitization for the performance data being processed.
[COMMAND_EXECUTION]: The skill's workflow involves executing local Node.js scripts and CLI tools, such as node tools/clis/google-ads.js, to fetch data and manage ad campaigns.
[EXTERNAL_DOWNLOADS]: The reference documentation suggests the use of external tools from third-party sources, including cloning the voicebox repository from GitHub and utilizing npx for Remotion rendering.

ad-creative