unified-web-search

What This Skill Does

The unified-web-search skill serves as a sophisticated aggregation engine for the OpenClaw AI agent, designed to synthesize information from diverse environments. Rather than relying on a single data source, this tool intelligently orchestrates requests across Tavily (for AI-optimized web results), Web Search Plus, raw browser scraping, and local workspace file indexing. It manages the complexities of cross-platform retrieval, normalizing the results into a single, ranked JSON output complete with provenance tracking. This allows the AI agent to maintain a clear context of where each piece of information originated, ensuring transparency and reliability in its reasoning processes.

Installation

You can install this skill directly via the OpenClaw registry using the following command in your terminal:

clawhub install openclaw/skills/skills/iyeque/unified-web-search

Ensure that you have your TAVILY_API_KEY configured in your environment variables, as the tool requires it for optimal web search performance. Once installed, the skill is immediately available for the AI to invoke based on your natural language prompts.

Use Cases

This skill is perfect for scenarios requiring a mix of broad internet context and specific local knowledge. For instance, developers can use it to find the latest documentation for a new library while simultaneously searching local project notes for internal configuration standards. It is also highly effective for market research, allowing the AI to scrape specific URLs via the browser tool while fetching general industry trends through Tavily.

Example Prompts

1. "Search for the latest security updates regarding the Log4j vulnerability and cross-reference them with our internal security-policy.md file in the project folder."
2. "Find recent trends in transformer model architectures using web search, and check my local folder for any summaries I saved about this topic last month."
3. "Look up the documentation for the latest Node.js release and identify if any current breaking changes affect our package.json setup."

Tips & Limitations

To get the best results, keep your queries concise and avoid shell-sensitive characters; the skill includes built-in sanitization to prevent injection attacks, but clean input yields better relevance. Note that while local file searching is powerful, it is strictly sandboxed to your workspace directories for security. If a search feels sluggish, consider limiting the sources parameter to only the specific platforms you need rather than querying all four. Always ensure your environment variables are correctly exported to avoid partial failures where only local data is returned.