bocha-search

What This Skill Does

The bocha-search skill for OpenClaw integrates the powerful Bocha AI Search API, providing your agents with real-time access to the entire web. It is specifically designed for Large Language Models (LLMs), returning structured data that is easy for AI to parse and synthesize. Whether you need the latest market insights, research on complex technical topics, or a quick fact-check, this tool delivers high-accuracy results, optional rich summaries, and intelligent filtering options.

Installation

To integrate this tool into your OpenClaw environment, ensure you have your API Key ready from the Bocha Open Platform. Execute the following command in your terminal:

clawhub install openclaw/skills/skills/sunjingji/bocha-search

After installation, configure your credentials by either creating a config.json file in the root directory or setting the BOCHA_API_KEY environment variable. The skill is then ready to be invoked directly via your AI workflows.

Use Cases

• Market Research: Quickly gather summaries and official reports on specific industries or public companies.
• Content Creation: Research trending topics or news events to provide context-aware drafts for blog posts or social media.
• Fact Checking: Retrieve authoritative, time-stamped sources to verify information claims within a chat session.
• Technical Documentation: Search for the latest library updates, documentation, or coding best practices across the web.

Example Prompts

1. "Search for the latest developments in DeepSeek AI and summarize the key technological breakthroughs from the last month."
2. "Find recent ESG reports for Alibaba and summarize the top three sustainability goals mentioned in the search results."
3. "What are the current performance benchmarks for the latest AI models? Use the search tool to find relevant articles and summarize them for me."

Tips & Limitations

To get the best results, use the default noLimit setting, as the Bocha search algorithm is highly capable of optimizing time parameters automatically. Using restrictive time filters like oneDay may occasionally return zero results if no relevant content was indexed in that specific window. If you are processing long-form content, always enable the --summary flag to save tokens and improve the relevance of the information extracted. Keep in mind that search results are subject to web index availability; if you encounter a 429 error, simply wait a few seconds before retrying to respect rate limits.