The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from rendered HTML files, sitemaps, and robots.txt files, which could contain hidden instructions designed to influence the agent's behavior.
Ingestion points: The skill is instructed in SKILL.md and references/technical.md to read and parse "HTML atual", "sitemap", and "robots".
Boundary markers: The instructions do not define clear delimiters or specify that the content of these files should be treated as untrusted data.
Capability inventory: The skill executes shell commands via the astrofy CLI and performs extensive file system read operations across the project structure.
Sanitization: No sanitization, validation, or escaping protocols are described for the content ingested from the audited files.

astrofy-seo