> ## Documentation Index
> Fetch the complete documentation index at: https://hyperbrowser.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> Use AI agents to easily automate browser tasks and create agentic workflows

Hyperbrowser lets you run powerful, AI-driven browser agents in managed cloud sessions. Whether you prefer open-source frameworks or cutting‑edge model-native agents, you can start tasks with a single API call and watch them execute live.

<Info>
  All agents share the same operational model: start a task, optionally poll for
  status, and fetch the final result. Use our SDK helpers like `startAndWait()`
  for a simple blocking workflow, or the async pattern for full control.
</Info>

## Which agent should I use?

<CardGroup cols={2}>
  <Card title="Browser-Use" icon="compass" href="/agents/browser-use" iconType="regular">
    Open-source, fast, and efficient framework for browser automation with
    strong defaults and great performance. Interacts with the dom using
    Playwright.
  </Card>

  <Card title="Claude Computer Use" icon="microchip-ai" href="/agents/claude-computer-use" iconType="regular">
    Anthropic’s Claude with full computer-use capabilities for robust, reliable
    real‑world interaction. Has wide array of actions and computer tools that it
    can choose from.
  </Card>

  <Card title="OpenAI CUA" icon="brain-circuit" href="/agents/openai-cua" iconType="regular">
    OpenAI’s Computer‑Using Agent (Operator tech) for task execution across the
    web UI. Usually slower but can be more reliable/accurate.
  </Card>

  <Card title="Gemini Computer Use" icon="sparkles" href="/agents/gemini-computer-use" iconType="regular">
    Google’s Gemini with computer-use for fast agentic tasks. Has more broad
    tools that can incorporate multiple actions into a single step.
  </Card>

  <Card title="HyperAgent" icon="bolt" href="/agents/hyperagent" iconType="regular">
    Our open-source, Playwright‑powered agent framework designed for control and
    extensibility.
  </Card>

  <Card title="Stagehand" icon="cursor-click" href="/integrations/stagehand" iconType="regular">
    Run Stagehand on Hyperbrowser-managed browsers. Attach via CDP, reuse
    stealth sessions, and monitor tasks live without self-hosting Chrome.
  </Card>
</CardGroup>

## Quickstart

The simplest way to run any agent is to call its `startAndWait()` method from our SDKs. Here’s a representative example using Browser‑Use:

<CodeGroup>
  ```typescript Node.js theme={null}
  import { Hyperbrowser } from "@hyperbrowser/sdk";
  import { config } from "dotenv";

  config();

  const client = new Hyperbrowser({
    apiKey: process.env.HYPERBROWSER_API_KEY,
  });

  async function main() {
    const result = await client.agents.browserUse.startAndWait({
      task: "Go to Hacker News and tell me the title of the top post",
      llm: "gemini-2.0-flash",
      maxSteps: 20,
    });

    console.log(`Output:\n${result.data?.finalResult}`);
  }

  main().catch((err) => {
    console.error(`Error: ${err.message}`);
  });
  ```

  ```python Python theme={null}
  from hyperbrowser import Hyperbrowser
  from hyperbrowser.models import StartBrowserUseTaskParams
  import os
  from dotenv import load_dotenv

  load_dotenv()

  client = Hyperbrowser(api_key=os.getenv("HYPERBROWSER_API_KEY"))

  result = client.agents.browser_use.start_and_wait(
      params=StartBrowserUseTaskParams(
          task="Go to Hacker News and tell me the title of the top post",
          llm="gemini-2.0-flash",
          max_steps=20
      )
  )

  print(f"Output:\n{result.data.final_result}")
  ```
</CodeGroup>

<Tip>
  Switch the agent family by swapping the SDK path, e.g.
  `client.agents.claudeComputerUse.startAndWait(...)`,
  `client.agents.cua.startAndWait(...)`,
  `client.agents.geminiComputerUse.startAndWait(...)`, or
  `client.agents.hyperAgent.startAndWait(...)`.
</Tip>

## Best practices

<AccordionGroup>
  <Accordion title="Write clear, specific tasks">
    Be explicit about the goal and constraints. Prefer “go to example.com, open
    pricing, extract Enterprise monthly price” to vague prompts.
  </Accordion>

  <Accordion title="Tune maxSteps and failure budgets">
    Simple tasks typically succeed within 10–20 steps; complex multi‑page flows
    may need 50+. Monitor failures and adjust `maxSteps` and `maxFailures`.
  </Accordion>

  <Accordion title="Reuse sessions for multi‑step workflows">
    Create a session once, pass `sessionId` to successive tasks, and set
    `keepBrowserOpen: true` where you need continuity.
  </Accordion>

  <Accordion title="Bring your own LLM keys when desired">
    Set `useCustomApiKeys: true` and provider your own API Keys to pass calls to
    your own organization.
  </Accordion>
</AccordionGroup>

<Note>
  For best results with AI Agents, it is highly recommended to keep the default
  screen configuration of 1280 x 720. The models can behave poorly with larger
  screen sizes.
</Note>

## Explore agents

* [Browser‑Use](/agents/browser-use)
* [Claude Computer Use](/agents/claude-computer-use)
* [OpenAI CUA](/agents/openai-cua)
* [Gemini Computer Use](/agents/gemini-computer-use)
* [HyperAgent](/agents/hyperagent)