Deep Search

What This Skill Does

Deep Search is a powerful, multi-tiered AI search routing engine designed specifically for the OpenClaw ecosystem. It leverages Perplexity AI's sophisticated sonar model suite to provide context-aware, research-grade answers. Unlike standard search tools, Deep Search automatically routes your queries through three distinct tiers: 'quick' for rapid, low-latency sonar responses, 'pro' for comprehensive research using sonar-pro, and 'deep' for complex analytical tasks utilizing sonar-reasoning-pro. By intelligently selecting the appropriate model based on query complexity, the skill ensures efficiency and precision. It further enhances search quality by supporting specific focus modes, allowing users to narrow down results from specialized sources such as academic databases, global news outlets, YouTube video transcripts, and Reddit community discussions.

Installation

To integrate Deep Search into your OpenClaw agent, ensure you have Python 3.10 or higher installed and the requests package available in your environment. You must first obtain a valid API key from Perplexity AI and set it as the PERPLEXITY_API_KEY environment variable in your system configuration. Once set, execute the installation command: clawhub install openclaw/skills/skills/aiwithabidi/agxntsix-deep-search. This command will pull the necessary scripts from the AgxntSix repository, making the tool ready for immediate use within your agent's task pipeline.

Use Cases

This skill is indispensable for researchers, developers, and data analysts who require high-fidelity information retrieval. It excels in tasks like literature reviews, competitive market analysis, tracking breaking news, and summarizing long-form media content. By using the 'deep' mode, users can offload complex reasoning chains to the agent, which handles synthesis and analysis automatically.

Example Prompts

1. "Deep Search, provide a detailed summary of the latest breakthroughs in transformer architecture from the last three months using academic sources."
2. "Research the comparative advantages of Python vs Go for high-concurrency microservices, then summarize the findings in a table."
3. "Search Reddit for community feedback on the latest OpenClaw release and list the top three commonly requested features."

Tips & Limitations

For best results, specify a focus mode whenever the query is domain-specific. Avoid using the 'deep' mode for trivial questions to conserve token usage and reduce latency. Be aware that this skill requires an active internet connection and a valid API subscription. Since the skill relies on an external API, performance is contingent on Perplexity's service availability.