The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze untrusted data from external domains (specifically robots.txt files and website headers).
Ingestion points: Instructions in SKILL.md ("Fetch /robots.txt on the live domain") and references/rule.md ("Fetch /robots.txt directly") require the agent to process content from attacker-controlled or third-party websites.
Boundary markers: The instructions do not provide explicit delimiters or "ignore embedded instructions" warnings for the agent when processing the fetched content.
Capability inventory: The skill uses network operations to fetch external files and expects the agent to review rendered HTML and response headers.
Sanitization: There is no mention of sanitizing or escaping the external content before the agent processes it.
[EXTERNAL_DOWNLOADS]: The skill performs standard HTTP requests to fetch public configuration files (robots.txt) for SEO auditing purposes. These operations target the production domain being audited.

robots-txt