The Agent Skills Directory

[SAFE]: The skill operates within a restricted tool surface using the Hyper MCP (firecrawl_extract_branding, openai_image_edit, etc.). All tool calls are aligned with the stated purpose of generating advertising assets.
[SAFE]: External website data is ingested only through the firecrawl_extract_branding tool, which is a standard method for brand analysis. The skill does not attempt to perform arbitrary network requests or exfiltrate sensitive local data.
[SAFE]: All external references, such as the Hyper platform (app.hyperfx.ai), are legitimate resources belonging to the skill's vendor (hyperfx-ai).
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an inherent attack surface by processing data from user-supplied URLs.
Ingestion points: External website content is processed via firecrawl_extract_branding in SKILL.md.
Boundary markers: The instructions do not explicitly implement delimiters or specific instructions to ignore malicious text embedded in the source website's branding metadata.
Capability inventory: The skill uses image generation tools (openai_image_edit, nano_banana_image_generation) and generates advertising copy based on the ingested data.
Sanitization: There is no explicit sanitization step described for the branding personality or description strings before they are used in prompts.

ad-creative-generation