The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process unstructured data from external websites, which inherently creates a surface for indirect prompt injection.
Ingestion points: External URLs fetched via urllib, requests, or playwright (referenced in SKILL.md, scripts/recon.py, and examples/explore.py).
Boundary markers: The skill explicitly identifies this risk in SKILL.md (Step 6, 0a) and references/scrapling.md (Concept 1), advising users to strip hidden elements before passing content to an LLM.
Capability inventory: The skill uses standard network libraries and browser automation tools to fetch content.
Sanitization: A dedicated sanitization function sanitize_for_llm is provided in references/scrapling.md to remove script, style, template, and hidden elements.
[COMMAND_EXECUTION]: The documentation provides instructions for standard development environment setup, such as installing browser drivers via playwright install and package management via pip install or git clone.
[EXTERNAL_DOWNLOADS]: The skill references and provides instructions for downloading established tools and libraries. This includes Microsoft's Webwright framework from its official GitHub repository and various well-known scraping and parsing libraries like scrapling, playwright-stealth, and readability-lxml. These are documented as standard implementation details for the scraping workflow.

website-scraping