The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted content from the web, which creates a surface for indirect prompt injection attacks. * Ingestion points: Web pages fetched via CLI commands or Python sessions. * Boundary markers: The documentation explicitly instructs agents to use the --ai-targeted flag to sanitize content and includes logic to strip hidden elements and comments. * Capability inventory: The skill possesses capabilities to write to the filesystem, perform network operations, and execute browser automation scripts. * Sanitization: The tool implements sanitization logic that strips CSS-hidden elements, aria-hidden elements, and zero-width characters during AI-targeted extraction.
[EXTERNAL_DOWNLOADS]: The skill requires downloading browser binaries and container images during the setup phase. * Evidence: The scrapling install command downloads browser dependencies for Playwright-based fetchers. Docker images are fetched from ghcr.io and Docker Hub.
[COMMAND_EXECUTION]: The skill provides a CLI for web extraction and browser management. * Evidence: Commands such as scrapling extract get and scrapling extract fetch allow for interaction with web resources and local filesystem output.

scrapling-official