google-search

The google-search skill provides a high-fidelity web retrieval interface by leveraging an actual headless Chrome instance via the chromux CLI. Unlike standard API-based web searching, this tool executes JavaScript, allowing it to render complex modern websites and bypass bot detection mechanisms that often block simple crawlers. It is designed for developers, researchers, and technical agents who require reliable, human-like access to web data. Whether you need to extract full article bodies, retrieve comments from developer forums, or perform deep site-specific indexing, this skill provides the necessary control over the browsing session. It is the preferred choice when standard snippets provided by conventional search APIs are insufficient or when dynamic content rendering is mandatory for a successful retrieval operation.

• Full Chrome rendering engine: executes JS to fetch content exactly as a user sees it in a browser.

• Advanced filtering: supports Google operators like site:, alongside time-based filters for day, week, month, or year.

• Data extraction: includes specialized support for extracting article bodies, author metadata, tags, and comment sections.

• Structured output: provides both human-readable text summaries and raw JSON exports for programmatic integration.

• Flexible enrichment: optional enrichment mode allows you to toggle between fast snippet-only results and deep page parsing.

• Use this for deep research tasks where full text content is needed for RAG (Retrieval-Augmented Generation) pipelines.

• Excellent for searching domain-specific documentation or coding platforms like dev.to, stackoverflow.com, or GitHub discussions.

• Requires the chromux CLI pre-installed and a headless Chrome instance to be active for optimal performance.

• Be mindful of enrichment latency; prefer the --no-enrich flag for rapid query tasks to save time and system resources.

• Handles site-specific search queries efficiently by injecting operators directly into the Google search request.