The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill performs network requests using curl to fetch standard SEO-related files (such as robots.txt, sitemap.xml, and llms.txt) and raw HTML from the deployed website URL for verification. This is the primary function of the audit and targets the user-specified project domain.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it processes untrusted data from external websites and repository content files, which are then used to influence report generation and proposed code fixes.
Ingestion points: Site responses fetched via curl (raw HTML) and repository content files analyzed in the Content and GEO rubric categories.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the ingested site content.
Capability inventory: The skill writes an SEO.md file, generates reports in a subfolder, and has the capability to modify repository source code.
Sanitization: No explicit sanitization of the external HTML or content files is described.
Mitigation: A robust human-in-the-loop guardrail is implemented in Step 6, which requires explicit user review and acceptance before any suggested code fixes are applied to the source files.

seo