The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches and analyzes external web content via scripts/fetch_page.py, creating a potential surface for indirect prompt injection if the targeted pages contain malicious instructions. The skill proactively mitigates this risk through explicit instructions in SKILL.md and references/skill-contract.md directing the agent to treat all scraped content as untrusted evidence and to ignore embedded instructions.
Ingestion points: The scripts/fetch_page.py utility retrieves HTML content (titles, descriptions, and headers) from user-specified URLs.
Boundary markers: The skill requires the agent to label findings with evidence types (e.g., measured, estimated), which helps distinguish external data from internal logic.
Capability inventory: The skill has network access for fetching pages and file-writing capabilities for report generation.
Sanitization: scripts/fetch_page.py performs basic sanitization by stripping HTML tags and unescaping entities before data processing.
[EXTERNAL_DOWNLOADS]: The skill references several well-known third-party marketing and SEO services (such as Google Search Console, Ahrefs, HubSpot, and Stripe) and mentions the reportlab library in scripts/generate_pdf_report.py. These references are to established, reputable services and libraries necessary for the skill's documented functionality.
[SAFE]: The network utility scripts/fetch_page.py includes robust SSRF protections, such as validating that hostnames resolve to global (public) IP addresses and preventing access to private, loopback, or reserved IP ranges.

seo-page