what is best practice to setup web search for openclaw?

[full forge]

I'd like both text based search such as github and openclaw website, and also to interact with websites graphically (follow directions to peform human like tasks on a website). I'm on a mac.

[graceful jasper]

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

[dawn dune]

For “text-based” web access vs “human-like” website interaction, OpenClaw really has two different surfaces that work best together:

1) Text-based search + reading pages (GitHub, docs, blogs): web_search + web_fetch

• web_search = finds links (needs an API key)
• web_fetch = downloads a page + extracts readable text/markdown (no key required)

Recommended setup (Brave Search)

Run the wizard (easiest on macOS too):

openclaw configure --section web

That stores your key in:

• tools.web.search.apiKey

Config looks like (from the FAQ):

{
  tools: {
    web: {
      search: { enabled: true, apiKey: "BRAVE_API_KEY_HERE", maxResults: 5 },
      fetch: { enabled: true }
    }
  }
}

If you use tool allow/deny policy, make sure these are allowed:

• web_search / web_fetch or group:web

Docs:

• https://docs.openclaw.ai/tools/web
• https://docs.openclaw.ai/help/faq (see “How do I enable web search and web fetch”)

Practical “best practice” queries for your use case

• OpenClaw docs: site:docs.openclaw.ai <your topic>
• GitHub issues/PRs: site:github.com/openclaw/openclaw <error message>

2) Graphical / interactive websites (clicking, typing, following directions): browser

This is for JS-heavy sites, logins, forms, “do the thing on the website”.

You have two good options on a Mac:

Option A (safer/isolated): OpenClaw-managed browser profile (openclaw)

This launches a dedicated, isolated automation browser/profile.

openclaw browser --browser-profile openclaw start
openclaw browser --browser-profile openclaw open https://example.com
openclaw browser --browser-profile openclaw snapshot

Docs: https://docs.openclaw.ai/tools/browser

Option B (most “human-like”): Chrome extension relay (chrome profile)

This lets the agent drive your existing Chrome tab (great when you need an already-logged-in session), but it’s also the highest-trust option.

Install the extension:

openclaw browser extension install
openclaw browser extension path

Then in Chrome:

• chrome://extensions → Developer mode → Load unpacked → select that path
• Click the OpenClaw Browser Relay toolbar icon on the tab you want to attach (badge shows “ON”)

Docs: https://docs.openclaw.ai/tools/chrome-extension

Best practice: use a separate Chrome profile for this (not your daily-driver profile), since attaching gives the agent access to whatever that tab can access.