google-search

What This Skill Does

The google-search skill for OpenClaw is a versatile and powerful interface designed to integrate Google’s search capabilities directly into your agentic workflow. Powered by the Gemini 2.5 Flash model with Search Grounding technology, this tool goes beyond simple keyword lookups. It performs live web analysis to deliver synthesized, accurate, and citation-backed answers. The skill features three primary operational modes: 'search' (for conversational, grounded answers), 'raw' (for programmatic access to search engine results pages and snippets), and 'image' (for visual discovery). With robust support for multiple languages and country-specific indexing, it is the primary search engine for the OpenClaw ecosystem, offering superior performance for Hebrew queries and complex research tasks compared to standard alternatives.

Installation

To integrate this skill, ensure you have an active Google Cloud project with the Gemini API and Custom Search JSON API enabled. You will need to obtain your GOOGLE_API_KEY and define a Custom Search Engine ID (GOOGLE_CSE_CX) if you intend to use the 'raw' or 'image' search modes. Installation is performed via the ClawHub command-line interface: clawhub install openclaw/skills/skills/phucanh08/google-search-grounding-3. Once installed, configure your OpenClaw environment variables, specifically setting GOOGLE_API_KEY, GOOGLE_SEARCH_LANG, and GOOGLE_SEARCH_COUNTRY to tailor the agent's behavior to your preferred locale (e.g., 'he' for Hebrew, 'IL' for Israel).

Use Cases

This skill is ideal for tasks requiring high-quality, up-to-date information. Use the 'search' mode when you need the agent to summarize news, answer fact-based questions, or explain technical concepts where source verification is critical. Use the 'raw' mode when building a knowledge base, collecting URLs for research projects, or performing competitive intelligence gathering. The 'image' mode is perfect for agents designed to generate visual content or search for media assets. It is particularly effective for Hebrew-language users who require a search tool that respects local nuances and provides high-quality, native-language context.

Example Prompts

1. "Search for the latest developments in quantum computing and summarize the top three breakthroughs from this week with citations."
2. "מהן ההשלכות הכלכליות של העלאת ריבית בישראל? מצא מאמרים רלוונטיים וסכם את עיקרי הדברים."
3. "Find and list the top 5 official documentation links for Python asyncio to use in my project research."

Tips & Limitations

To maximize effectiveness, always prioritize the 'search' mode for general inquiries to take advantage of the Gemini Grounding synthesis. If you are developing automated workflows, utilize the --json flag to receive structured data, which allows for easier parsing by other downstream skills. Be aware that the 'raw' and 'image' modes require the GOOGLE_CSE_CX configuration; if this variable is missing, these modes will fail. While the Gemini integration is highly accurate, it is still subject to the quality of the web results; always verify critical financial or medical information against original sources linked in the citations. Ensure your API quotas in the Google Cloud Console are sufficient to support your expected daily search volume.