Power your AI agent with real-time Google Search results and Gemini-powered grounding for accurate, cited answers.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install google-search-grounding-3
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install google-search-grounding-3 using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Google Search Grounding?
The Google Search Grounding skill is a robust integration designed for Openclaw Skills that leverages the power of Gemini 2.0 Flash alongside the Google Custom Search JSON API. It provides a superior alternative to standard web search tools by offering synthesized answers with verifiable citations, ensuring that your AI agent provides grounded and factual information. This skill is built to handle complex research tasks, from retrieving raw web snippets to performing high-resolution image searches.
Optimized for developer workflows, this skill prioritizes performance and accuracy. It features native support for multiple languages, with a particular emphasis on high-quality Hebrew processing, making it an essential component for international Openclaw Skills deployments. By integrating Gemini Search Grounding, the skill minimizes hallucinations and provides a transparent trail of sources for every answer generated.
Google Search Grounding Use Cases
- Synthesizing direct answers to complex questions with numbered citations for factual verification.
- Extracting raw URLs and metadata for deep-dive research and competitive analysis.
- Sourcing high-quality images and visual references via the Custom Search API.
- Managing multilingual search queries within Openclaw Skills, specifically optimized for Hebrew and Arabic.
- Providing real-time grounding for current events where training data might be outdated.
How Google Search Grounding Works
- The AI agent invokes the skill using the python-based CLI, passing a specific mode such as search, raw, or image.
- In search mode, the query is routed through Gemini 2.0 Flash with Search Grounding enabled to produce a synthesized response.
- In raw or image modes, the tool queries the Google Custom Search JSON API to retrieve links, snippets, or image metadata.
- The skill automatically handles environment-based configuration for language and country-specific results in Openclaw Skills.
- Error handling logic manages API rate limits (429 errors) with automatic retries and provides fallbacks if quotas are exceeded.
- The processed data is returned in either a human-readable text format or a structured JSON object for further programmatic action.
Google Search Grounding Setup
To integrate this search capability into your Openclaw Skills environment, follow these steps:
- Obtain a
GOOGLE_API_KEYfrom the Google Cloud Console and enable the Gemini API and Custom Search JSON API. - Run the provided installation script:
bash skills/google-search/install.sh
- Configure your environment variables in the Openclaw configuration:
{
"env": {
"GOOGLE_API_KEY": "YOUR_KEY_HERE",
"GOOGLE_SEARCH_LANG": "he",
"GOOGLE_SEARCH_COUNTRY": "IL"
}
}
- If using raw or image modes, ensure the
GOOGLE_CSE_CX(Custom Search Engine ID) is also defined.
Google Search Grounding Data Schema & Taxonomy
The skill organizes search data into structured formats to be easily consumed by Openclaw Skills:
| Component | Description | Format |
|---|---|---|
| Synthesized Answer | AI-generated response based on search results | Markdown Text |
| Citations | Numbered list of sources mapped to the answer | List |
| Raw Results | Collection of titles, snippets, and source URLs | Markdown Table/List |
| Image Data | URLs to hosted images and their source titles | JSON/List |
| Metadata | Confidence scores and grounding supports | JSON (optional) |
Google Search Grounding Advanced Features
- Smart retry logic for 429 rate limits to ensure reliability in Openclaw Skills.
- Support for structured JSON output mode for automated data pipelines.
- Dynamic language and country resolution via CLI flags or environment variables.
- Seamless fallback mechanisms to alternative search providers when Google quotas are reached.
- Integrated Hebrew and Arabic language optimization for middle-eastern market applications.
SKILL.md
Loading
Related Openclaw Skills
Synology Surveillance Station
A powerful API bridge to control Synology Surveillance Station cameras, snapshots, and recordings directly through your automation workflows.
Lyrion Music Server Control
A specialized integration to control Lyrion Music Server playback, volume, and library management via JSON-RPC API.
Bambu Lab 3D Printer Controller
A robust MQTT-based interface to monitor, control, and automate Bambu Lab 3D printers directly from your local environment.
PhotoCHAT CLI Search
A sophisticated command-line interface for searching local photo libraries using natural language queries and advanced semantic filtering.
Multi Search Engine
A comprehensive search integration for AI agents providing access to 17 global and domestic search engines without the need for API keys.
Web Search
Perform real-time web searches directly within your AI agent's workflow to access the latest internet data.
Featured*
