The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests content from external landing pages to inform its ad copy generation, which creates a surface for malicious instructions to influence the model.
Ingestion points: The generateAdCopy function in SKILL.md accepts a landingPageUrl and uses scrapeLandingPage to pull content into the agent's context.
Boundary markers: There are no explicit markers or instructions provided in the snippet to separate the untrusted scraped content from the agent's core generation instructions.
Capability inventory: The skill is restricted to text analysis and generation and does not exhibit high-risk capabilities such as arbitrary file writes or system command execution.
Sanitization: The provided code does not show any filtering or validation of the landing page content before it is processed by the AI generation engine.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to retrieve data from external web sources.
The implementation in SKILL.md includes scrapeLandingPage and analyzeCompetitorAds which fetch data from external URLs and search results to provide context for the ad variants.

ad-copy-generation