How to Use OpenClaw Tavily

Tavily is an OpenClaw web provider built for AI-friendly search and extraction. The docs show two roles: use it as the generic web_search provider, or call plugin tools tavily_search and tavily_extract when you need provider-specific controls. It is a good middle ground between simple search snippets and heavier scraping workflows.

30-second answer

Create a Tavily API key, set TAVILY_API_KEY or configure plugins.entries.tavily.config.webSearch, then set tools.web.search.provider to tavily. Use web_search for query and count. Use tavily_search for depth, topic, answer summaries, and domain filters. Use tavily_extract for specific URLs.

Search options

tavily_search accepts query, search_depth, topic, max_results, include_answer, time_range, include_domains, and exclude_domains. Basic depth is the default and faster. Advanced depth is slower but higher relevance. Topics include general, news, and finance, so the agent can ask for real-time updates without pretending every search source is the same.

Extraction options

tavily_extract accepts one to twenty URLs, optional query, extract_depth, chunks_per_source, and include_images. Use query plus chunks_per_source when you only need relevant chunks instead of full pages. Try basic extract first; fall back to advanced for JavaScript-rendered SPAs, dynamic content, or tables that basic extraction misses.

Configuration notes

Choosing Tavily in onboarding or openclaw configure --section web enables the bundled plugin automatically. The documented baseUrl is https://api.tavily.com. Store keys in config or the Gateway environment. Keep provider-specific config under plugins.entries.tavily.config.webSearch.

Choosing the right path

Use generic web_search when you only need normal results. Use tavily_search when domain filters or answer summaries matter. Use tavily_extract when you already have URLs. That separation keeps agents from overusing expensive or slow paths for simple lookups.

Playbook angle

The Playbook pattern is to treat Tavily as structured research plumbing. It is not a replacement for judgment; it is a cleaner way to feed the agent reliable search and extraction inputs.

Runbook checklist

Before you automate this, run one small acceptance test with harmless input. Confirm the tool is available to the right agent, the credential is loaded from config or environment, the output shape matches the workflow, and the failure message is understandable to a tired operator. If the feature touches money, public channels, logged-in browsers, host commands, or customer data, put a review step before the side effect. If it only reads data, still record the source and timestamp so future sessions do not treat stale context as fresh truth. Keep the first version narrow, then expand once the logs show the agent is choosing the right tool for the right reason. When the docs are incomplete, prefer a conservative sentence over a clever invented shortcut that future agents cannot reliably verify. Add one monitoring habit as well: after the first real run, check the transcript or logs for missing prerequisites, broad prompts, stale assumptions, and accidental side effects. Tighten the instruction while the failure is fresh. The best OpenClaw workflows improve in small, documented passes instead of one giant rewrite after something breaks in public. For SEO pages, that same discipline matters: do not promise hidden capabilities, paid-provider limits, or setup shortcuts unless the current docs say so. Trust compounds when the guide is accurate even in the boring operational edge cases that matter during real maintenance windows.

Operator note

How to Use OpenClaw Tavily works best when it is written into a small runbook instead of treated as a magic switch. Record who owns the workflow, which config keys are allowed, which credentials are required, what the agent may do without approval, and what counts as a failure. OpenClaw gives agents broad tools, but the reliable version is boring: one source of truth, one verification step, and one rollback path when a provider or channel behaves differently than expected.