The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill provides the ability to process untrusted external content, creating a vulnerability surface. 1. Ingestion points: WebView and page.load methods in SKILL.md accept arbitrary URLs and HTML strings. 2. Boundary markers: No delimiters or isolation instructions are present in the provided snippets. 3. Capability inventory: callJavaScript, snapshot, pdf, and webArchiveData methods (SKILL.md) allow extraction of data from the browser context. 4. Sanitization: The skill lacks examples of content validation or output sanitization.

swiftui-webkit