Tavily Search Pro

**Quick judgment**: Tavily Search Pro is a mature, feature-rich web search and content-retrieval skill well-suited as a general-purpose “eyes on the web” component for OpenClaw agents. It covers search, extraction, crawling, mapping, and deep research with citations, and the CLI is clearly documented and consistent. **Key strengths** - Broad functionality: unified interface for search (general/news/finance), extraction, crawl, map, and research. - Good agent fit: JSON output options and structured research results are ideal for automated pipelines. - Operational robustness: explicit handling of API key configuration, rate limits, network errors, and timeouts. **Main risks & limitations** - **External dependency**: All capabilities rely on the Tavily API; outages, API changes, or rate limits will directly impact reliability. - **Cost/credits**: Usage maps to credits per call/page/URL; intensive crawling or research can become expensive if not rate-limited by the agent. - **Content & compliance risk**: Crawling and extraction may touch sites with restrictive terms of service or sensitive content; implement allowlists/controls at the agent level. - **LLM answer quality**: Synthesized answers and research outputs can still contain inaccuracies or biased interpretations despite citations. **Recommended scenarios** - As the primary web search layer for agents that need fresh, citation-backed information (news, finance, tech, research topics). - For agents performing documentation analysis: crawl + extract on docs sites with focused instructions and path filters. - For building lightweight knowledge graphs or URL inventories via map and crawl before downstream processing. - For RAG-style pipelines that need high-quality page extraction (markdown/text) as input to vector stores or summarizers. Use this skill when you need a well-integrated, multi-purpose web intelligence layer and are comfortable depending on an external API with associated cost and availability trade-offs.