The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data from web pages using the WebFetch tool, creating a surface for indirect prompt injection.
Ingestion points: The skill processes external content through the PageSnapshot (specifically rendered_dom and raw_html) obtained from arbitrary URLs.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to treat the ingested HTML content as data only or to ignore potential instructions embedded within the page content.
Capability inventory: The skill allows the use of powerful tools including Bash and WebFetch. An attacker could attempt to embed malicious instructions in a web page's HTML to influence the agent's behavior when these tools are active.
Sanitization: The instructions do not specify any sanitization, filtering, or escaping of the HTML content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill documentation includes instructions to use the Bash tool to execute a local Node.js script (node scripts/check-freshness.mjs) for finding verification. While this is part of the skill's intended functionality, it involves the execution of code within the agent's environment.

seo-freshness