Playground

Use firstValidate before coding

Check the target site, runtime settings, and output shape first.

SurfacesBrowser, Fetch, and Crawl

Choose Playwright control, one-page extraction, or bounded crawl.

HandoffCopy reproducible code

Generated SDK and cURL use public API auth, not console-only behavior.

Workbench Surfaces

BrowserBrowser

Open the console Browser workbench for login, clicks, complex waits, and multi-step Playwright flows.

FetchFetch

Open the console Fetch workbench to render one URL and extract text, HTML, or screenshots.

CrawlCrawl

Open the console Crawl workbench for bounded same-site collection with include/exclude rules.

Agents are governed by a separate public readiness contract. Use Browser, Fetch, and Crawl for production workflows today, and see the Agents contract for readiness criteria.

Before running a workflow, review four runtime controls: profile state, timeout, recording, and whether the browser should stay alive for human inspection after the script finishes. These settings are included in the generated SDK and cURL examples, so the same run can be reproduced outside the dashboard.

Browser Workflow

Use Browser when you already know the script you want to run, or when you need full control over Playwright.

export default async function run({ page, log }) {
  await page.goto('https://example.com', { waitUntil: 'networkidle' });
  log('Title:', await page.title());
}

After the run starts, use the live preview, output, browser logs, network events, and metrics tabs to debug the session without leaving the page.

Fetch and Crawl

Fetch and Crawl both start from an explicit URL. The difference is scope: Fetch renders one page, while Crawl collects a bounded set of same-site pages.

Fetch focuses on output shape and content narrowing: text, HTML, screenshot, and include/exclude selectors. Crawl focuses on boundaries: start URL, max pages, sitemap or same-site link following. In both cases, review the summary before moving generated scripts or integration code into production.

Crawl is same-site by default. This prevents accidental broad crawls and keeps playground runs easy to reason about.

Integration Code

Every Playground surface exposes SDK and cURL examples. They use public API authentication only:

export HIVEBROWSER_API_KEY=hb_live_...

The generated examples do not include dashboard delegation headers or live-view tokens. Treat the generated script as application code: review it, keep secrets out of logs, and move stable workflows into your own repository.

History and Debugging

Fetch and Crawl return to their own workbench with compact result cards. Fetch highlights HTTP status, title, content blocks, and screenshots. Crawl highlights pages, same-site links, sitemap candidates, and screenshots. Confirm the result first, then open the full Browser workbench only when deeper inspection is needed.

Recent runs keep the session id, status, duration, output preview, and links back to session details. Use them to compare successful and failed attempts without digging through the full session table.

For failed runs, open session details to inspect request ids, runtime logs, replay artifacts, and browser/network events.