The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from external websites.
Ingestion points: The skill ingests raw_html and rendered_dom from external web pages via the WebFetch tool for SEO auditing in SKILL.md.
Boundary markers: There are no instructions to wrap external content in delimiters or to ignore instructions embedded within the fetched HTML.
Capability inventory: The skill utilizes Bash, WebFetch, Read, Grep, and Glob tools, providing a wide range of actions that could be targeted by an injection.
Sanitization: No sanitization, escaping, or validation of the external HTML content is specified before it is processed.
[COMMAND_EXECUTION]: The skill instructions mention using the Bash tool to perform render_diff assertions and other auditing tasks. Executing shell commands that may involve or be influenced by untrusted data from external web pages represents a potential command injection risk if the data is not properly escaped.

seo-rendering