How to Use OpenClaw Exa Search

Exa Search is OpenClaw’s fit for semantic and content-rich web research. The docs describe Exa as a web_search provider with neural, keyword, and hybrid search modes plus built-in content extraction. Instead of only returning titles and snippets, Exa can include highlights, extracted text, and summaries when configured through the contents option.

30-second answer

Set EXA_API_KEY in the Gateway environment or configure the Exa plugin, then set tools.web.search.provider to exa. Use query and count for basic searches. Add type for search mode and contents when the agent needs extracted text, highlights, or summaries in the result payload.

Search modes

Exa supports auto, neural, fast, deep, deep-reasoning, and instant. Auto lets Exa pick. Neural is useful for meaning-based discovery. Fast and instant are good for quick checks. Deep and deep-reasoning are better when relevance matters more than latency. Pick the lightest mode that answers the question.

Content extraction

The contents object can request full text, highlights, or summaries. The docs show text true, highlights with numSentences, and summary true. If no contents option is provided, Exa defaults to highlights so result descriptions have useful excerpts. Descriptions resolve from highlights first, then summary, then full text.

Time filters

Exa supports freshness plus date_after and date_before, but freshness and explicit date ranges cannot be combined. That is a good constraint. If you need “this week,” use freshness. If you need an exact campaign window, use date_after/date_before and skip freshness.

Operational notes

Exa can return up to 100 results per query subject to search-type limits, and results are cached for 15 minutes by default. Use higher counts when you will actually inspect or aggregate the results. For quick answers, fewer results are cheaper and easier for the agent to reason over.

Playbook angle

The Playbook pattern is to use Exa when semantic discovery matters. If the task is “find pages like this idea,” Exa shines. If it is “check a known URL,” web_fetch is simpler.

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 Exa Search 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.