OpenClaw

Add hosted screenshots to OpenClaw

Drop the Grabbit skill into your self-hosted OpenClaw agent, or add the hosted MCP server, for full-page captures, selector targeting, delayed rendering, and cookie-banner handling. Nothing to install on the box it runs on.

Get a test keySee the install
Setup reviewed against OpenClaw MCP documentation on September 29, 2026. Evidence: source-reviewed. Not executed end to end in this release.OpenClaw MCP documentation

Two ways to wire it in

Use the hosted MCP server if OpenClaw is talking MCP, or wrap the one-line CLI in a skill. Either way the capture runs on Grabbit, so a self-hosted deployment stays lean.

MCP config
// Point OpenClaw's MCP config at the hosted server
{
  "mcp": {
    "servers": {
      "grabbit": {
        "transport": "streamable-http",
        "url": "https://mcp.grabbit.live",
        "headers": { "Authorization": "Bearer sk_live_..." }
      }
    }
  }
}
Skill (wraps the CLI)
# Or wrap the one-line CLI in a skill (skills/grabbit/SKILL.md).
# The capture is a single command (no browser for OpenClaw to manage):
npx grabbit.live <url> --out shot.png   # → hosted URL + local file
npx grabbit.live <url> --test           # free placeholder, no credits

The gap a browsing agent leaves

A self-driven browser needs Chromium, system dependencies, timeout handling, and process cleanup. Grabbit moves that screenshot step to a hosted call with cookie-banner handling, full-page support, and an image URL OpenClaw can read, store, or act on.

Prefer raw HTTP for a custom skill? Same endpoint. See the screenshot API reference.

Hosted screenshots without a browser in the agent runtime

A local headless browser means installing Chromium, keeping its dependencies current, bounding memory use, and handling timeouts. Grabbit runs the screenshot step on hosted rendering infrastructure and returns an image URL.

Local headless browser

  • Needs Chromium installed in the agent's runtime
  • Cookie and consent handling belongs to your automation
  • You own the timeouts, retries, and scaling
  • Breaks in CI, sandboxes, and remote runners with no browser

Grabbit

  • One call, nothing to install or run
  • Dismisses cookie and consent banners automatically
  • Handles lazy-loaded, JavaScript-heavy, and full-page captures
  • Runs from CI or sandboxes without a local browser install

Grabbit accepts public pages and supports cookie-banner handling, delayed JavaScript, selector targeting, and full-page output. Target sites can still reject automated traffic, and private or internal addresses are blocked.

Flat pricing, built for agents that loop

When an agent captures a page on every step, per-call metering turns into anxiety. Grabbit is one flat rate: $50 per year for 25,000 captures ($0.002 each). Annual subscription credits expire at term end; separately purchased top-ups do not expire while your account remains active. Test-environment keys render placeholders for integration checks. A real page capture needs a live key and available credits.

See full pricingCompare providers

OpenClaw FAQ

Why add Grabbit to OpenClaw if it can already browse?
OpenClaw's built-in browsing handles navigation and extraction. Grabbit adds a hosted screenshot call with full-page output, delayed rendering, selector targeting, and automatic cookie-banner handling, without installing Chromium in the agent runtime.
How do I add Grabbit to OpenClaw?
Two ways. Add the hosted MCP server (https://mcp.grabbit.live) to OpenClaw's MCP config, or wrap the one-line CLI (npx grabbit.live <url>) in a skills/grabbit/SKILL.md. Both authenticate with a Bearer key; a free test key works.
Does OpenClaw need a headless browser for this?
No. Whether you use the MCP server or the CLI skill, the render happens on Grabbit. Nothing to install on the box OpenClaw runs on, which keeps a self-hosted deployment lean.
Can it screenshot any site?
No screenshot service captures every site. Grabbit accepts public URLs, handles cookie and consent banners, and can wait for JavaScript-heavy pages. Private and internal addresses are blocked by SSRF protection, and target sites can still reject automated traffic.
What does it cost?
A flat $0.002 per live capture on the $50/year plan. Annual-plan credits expire at term end, including when you cancel. Cancellation takes effect at term end. Separately purchased top-up credits do not expire while your account remains active. Neither resets monthly. Synchronous test requests return a free placeholder for wiring, not a live screenshot. Real captures require a live key and available credits; zero credits returns HTTP 402.

Give OpenClaw eyes in two minutes

Free wiring test, one install, then your agent can screenshot public pages with a live key.

Get a test keyAll agent integrations