The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes content from external URLs provided by the user, which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Processes data from user-supplied URLs via commands like /seo audit, /seo page, and /seo technical.
Capability inventory: The skill can execute local Python scripts (scripts/*.py), perform network requests (via curl examples and API integrations), and spawn multiple sub-agents.
Boundary markers: No specific delimiters or 'ignore' instructions for external content are defined in the orchestration logic.
Sanitization: No explicit sanitization of ingested web content is documented in the provided skill files.
[COMMAND_EXECUTION]: The skill invokes local Python scripts to handle authentication, drift history, and report generation.
Evidence: Mentions python scripts/google_auth.py, python scripts/backlinks_auth.py, python scripts/drift_history.py, and python scripts/google_report.py.
Extension Installation: Includes installation scripts for the firecrawl extension (extensions/firecrawl/install.sh and install.ps1).
[EXTERNAL_DOWNLOADS]: Integrates with numerous external SEO services to retrieve ranking, backlink, and technical data.
Trusted/Well-known Services: References Google Search Console, PageSpeed Insights, Moz API, Bing Webmaster Tools, DataForSEO, Overpass API, Geoapify, and Nominatim. These are presented as standard data sources for the skill's primary functionality.

seo