---
title: Lightcone Documentation | Lightcone
description: One API to get any piece of work done. Build AI agents that see, think, and act on browsers and desktops.
---

Lightcone is a cloud platform for browser automation and AI agents. Create sandboxed browser sessions, run autonomous agents that see and act on web pages, or build your own computer-use agent (CUA) loop — all through a single API. No browser infrastructure to manage.

## Get started

|                                                                 |                                                              |
| --------------------------------------------------------------- | ------------------------------------------------------------ |
| [**Quickstart**](/guides/quickstart/index.md)                   | Go from zero to a working browser session in under 5 minutes |
| [**Authentication**](/guides/authentication/index.md)           | Get your API key and configure the SDK                       |
| [**How Lightcone works**](/guides/how-lightcone-works/index.md) | Architecture, session lifecycle, and the three API layers    |
| [**API Reference**](/api/index.md)                              | Full reference for all 46 endpoints                          |

## Choose your path

### I want to automate browsers programmatically

Create a cloud browser session and control it with clicks, typing, navigation, and screenshots. Full Chromium with stealth mode and proxy support.

- [Computers overview](/guides/computers/index.md) — sessions, lifecycle, and all available actions
- [Automate a browser](/guides/automate-a-browser/index.md) — end-to-end walkthrough
- [Manage browser tabs](/guides/browser-tabs/index.md) — multi-tab workflows

### I want AI agents to complete tasks for me

Give an instruction in natural language and let an autonomous agent drive a browser to complete it. Stream progress in real time.

- [Agent Tasks](/guides/agent-tasks/index.md) — start, monitor, pause, resume agents
- [Run an AI agent](/guides/run-an-agent/index.md) — streaming, polling, and steering

### I want to build my own computer-use agent

Use the OpenAI-compatible Responses API to build a multi-turn computer-use agent (CUA) loop. You control the agent at every step: the model decides what to click, you execute it, and feed back a screenshot.

- [Responses API](/guides/responses-api/index.md) — create responses, process actions, chain turns
- [Computer-use agent loop](/guides/cua-protocol/index.md) — full implementation guide

### I want text generation / Chat Completions

OpenAI-compatible Chat Completions with tool calling, streaming, and structured outputs. Drop-in replacement — just change the base URL.

- [Chat Completions](/guides/chat-completions/index.md) — usage, streaming, tool calling, models

## Real-world use cases

| Use case                                                             | What it does                                                                   |
| -------------------------------------------------------------------- | ------------------------------------------------------------------------------ |
| [**Web scraping**](/use-cases/web-scraping/index.md)                 | Extract structured data from any website, even behind logins and bot detection |
| [**Form automation**](/use-cases/form-automation/index.md)           | Fill and submit multi-step forms across legacy web apps without APIs           |
| [**Dashboard monitoring**](/use-cases/dashboard-monitoring/index.md) | Watch web dashboards for changes and trigger alerts                            |

## Integrations

Lightcone works with the tools you already use. Connect to agent frameworks, orchestration platforms, and AI providers.

|                                               |                                                   |                                                             |
| --------------------------------------------- | ------------------------------------------------- | ----------------------------------------------------------- |
| [LangChain](/integrations/langchain/index.md) | [OpenAI CUA](/integrations/openai-cua/index.md)   | [Anthropic Claude](/integrations/anthropic-claude/index.md) |
| [CrewAI](/integrations/crewai/index.md)       | [Browser-Use](/integrations/browser-use/index.md) | [Vercel AI SDK](/integrations/vercel-ai/index.md)           |
| [MCP Server](/integrations/mcp/index.md)      | [Playwright](/integrations/playwright/index.md)   | [Mastra](/integrations/mastra/index.md)                     |

[View all integrations](/integrations/overview/index.md)

## Install the SDK

Terminal window

```
pip install tzafon
```

Terminal window

```
npm install @tzafon/lightcone
```

Having trouble? See the [Troubleshooting guide](/guides/troubleshooting/index.md).