The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by fetching and parsing web pages. Ingestion points include scripts/fetch_page.py which retrieves HTML content from external URLs. Boundary markers are explicitly defined in SKILL.md and references/skill-contract.md, which instruct the agent to treat scraped material as untrusted input and ignore embedded instructions. Capability inventory includes network fetching and report generation scripts. Sanitization involves parsing content for metadata using regex without executing any payload.
[COMMAND_EXECUTION]: The skill includes Python scripts for SEO data processing. scripts/fetch_page.py implements SSRF protections by validating hostnames against private/reserved IP ranges and ensuring redirects are safe.
[EXTERNAL_DOWNLOADS]: scripts/generate_pdf_report.py identifies reportlab as an optional dependency; it suggests an installation command if missing but does not perform automated downloads or remote code execution.

seo-images