x402agent/SolanaOS / skills/cua
cua
VerifiedComputer Use Agent β autonomous browser control via Steel, Browser Use, and Browserbase cloud providers. Supports headful CDP sessions, CAPTCHA solving, proxy rotation, live view, session recording/replay, Playwright/Puppeteer integration, and LLM-driven visual browsing. Use when asked about computer use, browser automation, web scraping, Steel sessions, cloud browsers, headful browsing, CAPTCHA solving, or autonomous web navigation.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 14.4 KB |
SKILL.md
View on GitHub---
name: cua
description: "Computer Use Agent β autonomous browser control via Steel, Browser Use, and Browserbase cloud providers. Supports headful CDP sessions, CAPTCHA solving, proxy rotation, live view, session recording/replay, Playwright/Puppeteer integration, and LLM-driven visual browsing. Use when asked about computer use, browser automation, web scraping, Steel sessions, cloud browsers, headful browsing, CAPTCHA solving, or autonomous web navigation."
version: 1.0.0
emoji: "π€"
tags:
- cua
- computer-use
- steel
- browseruse
- browserbase
- playwright
- automation
- browser
- scraping
- captcha
requires:
env:
- STEEL_API_KEY
- BROWSERUSE_API_KEY
- BROWSERBASE_API_KEY
- BROWSERBASE_PROJECT_ID
bins: []
allowed-tools:
- cua
- steel
- browse
- browseruse
- desktop
---
# CUA β Computer Use Agent
Autonomous browser control powered by three cloud browser providers: **Steel**, **Browser Use**, and **Browserbase**. Each offers CDP-based headful browser sessions for Playwright/Puppeteer automation, LLM-driven visual browsing, and human-in-the-loop workflows.
## Quick Start
```bash
# Environment setup
export STEEL_API_KEY="ste-..."
export BROWSERUSE_API_KEY="bu_..."
export BROWSERBASE_API_KEY="bb_live_..."
export BROWSERBASE_PROJECT_ID="446f4b4c-..."
# Daemon commands
/cua new # Create a Steel session (default provider)
/cua new --provider browserbase # Create a Browserbase session
/cua new --provider browseruse # Create a Browser Use session
/cua browse https://example.com # Navigate in active session
/cua screenshot # Capture current page
/cua status # Show session details + live view URL
/cua release # End session
/cua list # List all active sessions
```
## Architecture
```
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β SolanaOS CUA Skill β
β β
β ββββββββββββββββ βββββββββββββββββ βββββββββββββββββββββββββ β
β β Steel β β Browser Use β β Browserbase β β
β β Provider β β Provider β β Provider β β
β β β β β β β β
β β β’ Sessions β β β’ CDP URL β β β’ Sessions β β
β β β’ Proxy β β β’ Cloud β β β’ Extensions β β
β β β’ CAPTCHA β β β’ Profiles β β β’ Stealth Mode β β
β β β’ Live View β β β’ Proxy β β β’ Proxies β β
β β β’ Recording β β β β β’ Contexts β β
β β β’ Mobile β β β β β’ Functions β β
β β β’ Regions β β β β β’ Live View β β
β β β’ HLS Replay β β β β β’ Session Recording β β
β ββββββββ¬ββββββββ ββββββββ¬βββββββββ ββββββββββββ¬βββββββββββββ β
β β β β β
β ββββββββ΄βββββββββββββββββββ΄ββββββββββββββββββββββββ΄ββββββββββββ β
β β CDP (Chrome DevTools Protocol) β β
β β WebSocket Connection β β
β ββββββββββββββββββββββββββββ¬βββββββββββββββββββββββββββββββββββ β
β β β
β ββββββββββββββββββββββββββββ΄βββββββββββββββββββββββββββββββββββ β
β β Playwright / Puppeteer / Selenium β β
β β β β
β β β’ Page navigation β’ Form automation β β
β β β’ Element interaction β’ Screenshot capture β β
β β β’ File download/upload β’ PDF generation β β
β β β’ Network interception β’ Cookie management β β
β ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ β
β β
β ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ β
β β LLM Vision Agent (Optional) β β
β β β β
β β Screenshot β Vision LLM β Action Plan β Playwright Exec β β
β β β β
β β Supported: Claude, GPT-4o, Gemini, Llama 3.2, Qwen 2.5 VL β β
β ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ β
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
```
## Provider Comparison
| Feature | Steel | Browser Use | Browserbase |
|---------|-------|-------------|-------------|
| CDP WebSocket | β
| β
| β
|
| Playwright | β
| β
| β
|
| Puppeteer | β
| β
| β
|
| Selenium | β
| β | β
|
| Live View | β
WebRTC | β | β
iframe |
| Session Recording | β
HLS/MP4 | β | β
rrweb |
| Proxy | β
Residential | β
| β
|
| CAPTCHA Solving | β
Auto | β | β
Auto |
| Stealth Mode | β
Fingerprint | β | β
Advanced |
| Mobile Mode | β
| β | β
Viewport |
| Multi-Region | β
LAX/ORD/IAD | β | β
Multiple |
| Extensions | β | β | β
|
| Contexts (persist state) | β | β
Profiles | β
|
| Keep Alive | β | β | β
|
| Functions (serverless) | β | β | β
|
| Max Timeout | 24h | N/A | Plan-based |
| Human-in-the-loop | β
interactive | β | β
interactive |
## Steel Integration
### Creating Sessions
```go
// One-line Playwright connection (simplest)
browser := chromium.ConnectOverCDP("wss://connect.steel.dev?apiKey=YOUR_KEY")
// Or with full options
session := steel.Create(ctx, &steel.CreateOptions{
UseProxy: true,
SolveCaptcha: true,
Timeout: 600000, // 10 minutes
Region: "lax",
DeviceConfig: &steel.DeviceConfig{Device: "mobile"},
})
```
### Session Lifecycle
```
CREATE β LIVE β [automation] β RELEASE
β
FAILED (crash/timeout)
```
- Sessions auto-release after timeout (default 5 min)
- CDP inactivity timeout: 10 min without commands
- Send heartbeat every 5 min to keep alive: `page.evaluate(() => undefined)`
- Always call `Release()` when done
### Live View & Recording
```go
// Embed live view (read-only)
iframe := session.DebugURL + "?interactive=false"
// Embed with control (human-in-the-loop)
iframe := session.DebugURL + "?interactive=true&showControls=true"
// Get HLS recording playlist
playlist := steel.HLSPlaylist(session.ID)
// Use with hls.js or native Safari HLS
```
### Available Regions
| Region | Code | Location |
|--------|------|----------|
| Los Angeles | `lax` | Los Angeles, USA |
| Chicago | `ord` | Chicago, USA |
| Washington DC | `iad` | Washington DC, USA |
## Browserbase Integration
### Creating Sessions
```go
// Via Browserbase API
session := browserbase.CreateSession(ctx, cfg, "default")
// Returns CDP URL for Playwright connection
browser := chromium.ConnectOverCDP(session.CDPURL)
context := browser.Contexts()[0]
page := context.Pages()[0]
```
### Features
- **Stealth Mode**: Automatic fingerprinting, advanced stealth (Scale plan)
- **Proxies**: Built-in residential proxy network
- **Contexts**: Persist cookies/storage across sessions
- **Extensions**: Load custom Chrome extensions
- **Functions**: Deploy automation as serverless API endpoints
- **Search**: Web search without browser session
- **Fetch**: HTTP fetch without browser session
### Session Inspector
Every Browserbase session is recorded and can be inspected:
- Live browser state and interactions
- Network requests/responses
- Console output and errors
- Performance metrics
## Browser Use Integration
### Cloud Sessions
```go
// Create cloud browser via API
session := browserUseProvider.CreateSession(ctx, cfg, "default")
// Returns CDP URL
// Or via CLI
browser-use cloud connect
```
### Session Management
```go
// List sessions
browser-use sessions
// Close specific session
browser-use --session mySession close
// Close all sessions
browser-use close --all
```
## Daemon Commands
| Command | Description |
|---------|-------------|
| `/cua new [--provider steel\|browserbase\|browseruse]` | Create session |
| `/cua browse <url>` | Navigate to URL (via shell Playwright script) |
| `/cua screenshot` | Capture current page |
| `/cua status` | Show active session details |
| `/cua release` | End current session |
| `/cua release --all` | End all sessions |
| `/cua list` | List active sessions |
| `/cua live` | Get live view URL |
| `/cua replay <session-id>` | Get HLS replay URL |
| `/cua agent <objective>` | Start autonomous CUA agent |
## Autonomous CUA Agent
The CUA agent combines cloud browser sessions with vision LLMs for autonomous web navigation.
### Agent Loop
```
1. CREATE SESSION β Steel/Browserbase/BrowserUse
2. SCREENSHOT β Capture page via CDP
3. VISION ANALYSIS β Send to Claude/GPT-4o/Gemini
4. ACTION PLANNING β LLM determines next step
5. PLAYWRIGHT EXEC β Execute action via CDP
6. REPEAT β Until objective complete or max steps
7. RELEASE β Clean up session
```
### Vision Prompt Template
```
You are a Computer Use Agent controlling a cloud browser via Playwright.
OBJECTIVE: {objective}
Current page screenshot is attached. Analyze the page and determine the next action.
Available actions:
- page.goto(url) β Navigate to URL
- page.click(selector) β Click element
- page.fill(selector, text) β Fill input
- page.press(selector, key) β Press key
- page.screenshot() β Capture page
- page.waitForSelector(selector) β Wait for element
- page.evaluate(js) β Run JavaScript
- DONE β Task is complete
Respond with JSON: {"action": "...", "args": {...}, "reasoning": "..."}
```
## Environment Variables
| Variable | Provider | Required | Description |
|----------|----------|----------|-------------|
| `STEEL_API_KEY` | Steel | For Steel | Steel API key |
| `BROWSERUSE_API_KEY` | Browser Use | For BU | Browser Use cloud API key |
| `BROWSERBASE_API_KEY` | Browserbase | For BB | Browserbase API key |
| `BROWSERBASE_PROJECT_ID` | Browserbase | For BB | Browserbase project ID |
| `BROWSERBASE_PROXIES` | Browserbase | No | Enable proxy (default: true) |
| `BROWSERBASE_ADVANCED_STEALTH` | Browserbase | No | Advanced stealth mode |
| `BROWSERBASE_KEEP_ALIVE` | Browserbase | No | Keep alive after disconnect |
| `BROWSERBASE_SESSION_TIMEOUT` | Browserbase | No | Session timeout (ms) |
| `ANTHROPIC_API_KEY` | Vision LLM | No | For Claude vision agent |
| `OPENAI_API_KEY` | Vision LLM | No | For GPT-4o vision agent |
## API Reference
### Steel API
```
POST /v1/sessions Create session
GET /v1/sessions List sessions
GET /v1/sessions/:id Get session details
POST /v1/sessions/:id/release Release session
POST /v1/sessions/release Release all sessions
GET /v1/sessions/:id/screenshot Capture screenshot
GET /v1/sessions/:id/events Get replay events
GET /v1/sessions/:id/hls Get HLS playlist
WSS connect.steel.dev?apiKey=...&sessionId=... CDP WebSocket
```
### Browserbase API
```
POST /v1/sessions Create session
GET /v1/sessions/:id Get session details
POST /v1/sessions/:id Close session (REQUEST_RELEASE)
GET /v1/sessions/:id/debug Get debug URLs
```
### Browser Use API
```
POST /api/v2/browsers Create cloud browser
PATCH /api/v2/browsers/:id Stop browser (action: "stop")
```
## Playwright Connection Patterns
### Steel (one-liner)
```typescript
import { chromium } from "playwright-core";
const browser = await chromium.connectOverCDP(
`wss://connect.steel.dev?apiKey=${process.env.STEEL_API_KEY}`
);
```
### Steel (managed session)
```typescript
import Steel from "steel-sdk";
import { chromium } from "playwright";
const client = new Steel({ steelAPIKey: process.env.STEEL_API_KEY });
const session = await client.sessions.create({
useProxy: true,
solveCaptcha: true,
});
const browser = await chromium.connectOverCDP(
`wss://connect.steel.dev?apiKey=${process.env.STEEL_API_KEY}&sessionId=${session.id}`
);
const page = browser.contexts()[0].pages()[0];
await page.goto("https://example.com");
await browser.close();
await client.sessions.release(session.id);
```
### Browserbase
```typescript
import { chromium } from "playwright-core";
import { Browserbase } from "@browserbasehq/sdk";
const bb = new Browserbase({ apiKey: process.env.BROWSERBASE_API_KEY });
const session = await bb.sessions.create();
const browser = await chromium.connectOverCDP(session.connectUrl);
const page = browser.contexts()[0].pages()[0];
await page.goto("https://example.com");
```
### Browser Use (CDP)
```typescript
import { chromium } from "playwright-core";
// After creating a session via API, connect with CDP URL
const browser = await chromium.connectOverCDP(session.cdpUrl);
```
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |