### Install and Run agent-browser Demo

Source: https://github.com/vercel-labs/agent-browser/blob/main/examples/environments/README.md

Install dependencies and start the local development server for the agent-browser environments demo.

```bash
cd examples/environments
pnpm install
pnpm dev
```

--------------------------------

### Quick Start (No Install)

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/installation/page.mdx

Run Agent Browser commands directly using npx without a prior installation. The 'install' command downloads Chrome if it's the first time.

```bash
npx agent-browser install   # Download Chrome (first time only)
npx agent-browser open example.com
```

--------------------------------

### Install Dependencies and Setup Environment

Source: https://github.com/vercel-labs/agent-browser/blob/main/benchmarks/README.md

Installs project dependencies using pnpm and copies the environment configuration file. Ensure you have Node.js 18+ and pnpm installed.

```bash
cd benchmarks
pnpm install
cp .env.example .env
```

--------------------------------

### Install from Source

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/installation/page.mdx

Clones the repository, installs dependencies, builds the project, and links it globally. The 'install' command downloads Chrome on the first use.

```bash
git clone https://github.com/vercel-labs/agent-browser
cd agent-browser
pnpm install
pnpm build
pnpm build:native
./bin/agent-browser install
pnpm link --global
```

--------------------------------

### Install and Open URL with Agent Browser CLI

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/SKILL.md

Installs the agent-browser globally and opens a specified URL. Run 'agent-browser install' once after global installation.

```bash
# Install once
npm i -g agent-browser && agent-browser install

# Take a screenshot of a page
agent-browser open https://example.com
agent-browser screenshot home.png
agent-browser close
```

--------------------------------

### Start Recording with Initial Navigation

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/recording/page.mdx

Initiate a browser session and immediately start recording to a WebM file, navigating to a specified URL.

```bash
agent-browser open
agent-browser record start ./demo.webm https://example.com
```

--------------------------------

### Install and Get Core Skill

Source: https://github.com/vercel-labs/agent-browser/blob/main/skills/agent-browser/SKILL.md

Install the agent-browser CLI globally and then fetch the core skill content for workflows and troubleshooting. This is the starting point for using the tool.

```bash
npm i -g agent-browser && agent-browser install
```

```bash
agent-browser skills get core
```

```bash
agent-browser skills get core --full
```

--------------------------------

### Start and Open Session

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/dashboard/page.mdx

Start the dashboard server and open a session to a website.

```bash
agent-browser dashboard start
agent-browser open example.com
```

--------------------------------

### Start MCP Server with All Tools

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/commands.md

Starts the MCP server including all available tools for full CLI parity and advanced features.

```bash
agent-browser mcp --tools all
```

--------------------------------

### Start Tracing

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/commands/page.mdx

Command to start the tracing process.

```bash
agent-browser trace start             # Start trace
```

--------------------------------

### Install Agent Browser and Run Commands

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/changelog/page.mdx

Install the agent-browser globally using npm and then use it to download Chrome, open a website, or take a snapshot.

```bash
npm install -g agent-browser      # 7 MB install
agent-browser install              # download Chrome
agent-browser open example.com
agent-browser snapshot
```

--------------------------------

### Start Video Recording

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/commands/page.mdx

Command to start recording video to a WebM file.

```bash
agent-browser record start <path>     # Start video recording (WebM)
```

--------------------------------

### Install and Manage Agent Browser

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Commands for installing, upgrading, and diagnosing the agent-browser installation. Use `--with-deps` on Linux to also install system dependencies.

```bash
agent-browser install                 # Download Chrome from Chrome for Testing (Google's official automation channel)
agent-browser install --with-deps     # Also install system deps (Linux)
agent-browser upgrade                 # Upgrade agent-browser to the latest version
agent-browser doctor                  # Diagnose the install and auto-clean stale daemon files
agent-browser doctor --fix            # Also run destructive repairs (reinstall Chrome, purge old state, ...)
agent-browser doctor --offline --quick  # Skip network probes and the live launch test
agent-browser mcp                     # Start an MCP stdio server
```

--------------------------------

### Start Profiling

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/commands/page.mdx

Command to start Chrome DevTools profiling.

```bash
agent-browser profiler start          # Start Chrome DevTools profiling
```

--------------------------------

### Start MCP Server with Default Tools

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/commands.md

Starts the Model Context Protocol server with the default 'core' tool profile for basic browser automation.

```bash
agent-browser mcp
```

--------------------------------

### Pre-navigation Setup (Batch Command)

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/commands.md

Execute a sequence of commands in a single turn to set up the browser environment before the first navigation.

```APIDOC
## Pre-navigation Setup (Batch Command)

### Description
Execute a sequence of commands in a single turn to set up the browser environment before the first navigation. This is useful for configuring network routes, cookies, or initial scripts before the initial page load.

### Command

- **batch**
  - Description: Executes an array of commands in sequence. Useful for staging browser state like network interception, cookies, or initial scripts before the first navigation.
  - Usage: `agent-browser batch [command1] [command2] ...`
  - Example:
    ```bash
    agent-browser batch \
      '["open"]' \
      '["network","route","*","--abort","--resource-type","script"]' \
      '["cookies","set","--curl","cookies.curl","--domain","localhost"]' \
      '["navigate","http://localhost:3000/target"]'
    ```

### Notes
Using `open` without a URL allows for a clean launch. Any interceptions, cookies, or init scripts registered will take effect on the *first* real navigation. This is beneficial for SSR-only debugging, protected-origin authentication, or capturing fresh state without interference from previous pages.
```

--------------------------------

### CLI Command: profiler start

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/profiler/page.mdx

Starts the performance profiling process for the browser session.

```APIDOC
## CLI Command: profiler start

### Description
Starts recording a performance profile for the current browser session.

### Parameters
#### Query Parameters
- **--categories** (string) - Optional - A comma-separated list of Chrome trace categories to capture (e.g., "devtools.timeline,v8.execute").

### Request Example
agent-browser profiler start --categories "devtools.timeline,v8.execute"
```

--------------------------------

### Install, Start, and Stop Agent Browser Dashboard

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/changelog/page.mdx

Commands to manage the local observability dashboard. The dashboard provides a live UI for browser sessions.

```bash
agent-browser dashboard install
agent-browser dashboard start
agent-browser open example.com    # session appears in dashboard
agent-browser dashboard stop
```

--------------------------------

### Install Appium and XCUITest Driver

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/ios/page.mdx

Install Appium globally and then install the XCUITest driver for iOS automation.

```bash
npm install -g appium
```

```bash
appium driver install xcuitest
```

--------------------------------

### Start, Perform Actions, and Stop Profiling

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/profiler/page.mdx

This sequence demonstrates the basic workflow for starting the profiler, performing browser actions, and then stopping the profiler to save the trace data to a file.

```bash
agent-browser profiler start
```

```bash
agent-browser navigate https://example.com
```

```bash
agent-browser click "#button"
```

```bash
agent-browser profiler stop ./trace.json
```

--------------------------------

### Start MCP Server with Specific Tools

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/commands.md

Starts the MCP server with a custom selection of tool profiles, such as 'core', 'network', and 'react'.

```bash
agent-browser mcp --tools core,network,react
```

--------------------------------

### Basic Recording Workflow

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/recording/page.mdx

Start a browser session, begin recording to a WebM file, perform actions, and then stop the recording.

```bash
agent-browser open https://example.com
agent-browser record start ./demo.webm
agent-browser snapshot -i
agent-browser click @e1

agent-browser record stop
```

--------------------------------

### Global Installation

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/installation/page.mdx

Installs Agent Browser globally using npm. Run 'agent-browser install' to download Chrome for Testing on the first use.

```bash
npm install -g agent-browser
agent-browser install  # Download Chrome from Chrome for Testing (first time)
```

--------------------------------

### Homebrew Installation (macOS)

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/installation/page.mdx

Installs Agent Browser on macOS using Homebrew. Run 'agent-browser install' to download Chrome on the first use.

```bash
brew install agent-browser
agent-browser install  # Download Chrome (first time)
```

--------------------------------

### Basic Video Recording Workflow

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/video-recording.md

Launch the browser, start recording to a WebM file, perform actions, and then stop the recording.

```bash
# Launch the browser, then start recording
agent-browser open https://example.com
agent-browser record start ./demo.webm

# Perform actions
agent-browser snapshot -i
agent-browser click @e1
agent-browser fill @e2 "test input"

# Stop and save
agent-browser record stop
```

--------------------------------

### Start MCP Server with Specific Tools

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Starts an MCP server with a custom set of tools. Tools can be specified individually or as 'all' for full CLI parity. Profiles can be combined using commas.

```bash
agent-browser mcp --tools all
agent-browser mcp --tools core,network,react
```

--------------------------------

### Snapshot Output Format Example

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/snapshot-refs.md

This is an example of the output format from the `snapshot` command, showing the hierarchical structure of elements with assigned refs.

```text
Page: Example Site - Home
URL: https://example.com

@e1 [header]
  @e2 [nav]
    @e3 [a] "Home"
    @e4 [a] "Products"
    @e5 [a] "About"
  @e6 [button] "Sign In"

@e7 [main]
  @e8 [h1] "Welcome"
  @e9 [form]
    @e10 [input type="email"] placeholder="Email"
    @e11 [input type="password"] placeholder="Password"
    @e12 [button type="submit"] "Log In"

@e13 [footer]
  @e14 [a] "Privacy Policy"
```

--------------------------------

### Snapshot Command Output Example

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/snapshots/page.mdx

Example of the default text output format for the snapshot command when using the -i option.

```bash
agent-browser snapshot -i
# Output:
# @e1 [heading] "Example Domain" [level=1]
# @e2 [button] "Submit"
# @e3 [input type="email"] placeholder="Email"
# @e4 [link] "Learn more"
```

--------------------------------

### Install agent-browser CLI

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/page.mdx

Install the agent-browser CLI globally using npm or Homebrew. Download Chrome if it's the first time using the tool.

```bash
npm install -g agent-browser      # all platforms
brew install agent-browser        # macOS
agent-browser install             # Download Chrome (first time)
```

--------------------------------

### AgentCore Configuration Examples

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/providers/agentcore/page.mdx

Common usage patterns including persistent profiles, region selection, and SSO profile authentication.

```bash
# Basic usage (credentials auto-resolved via AWS CLI)
agent-browser -p agentcore open https://example.com

# With a browser profile for persistent login state
AGENTCORE_PROFILE_ID=my-profile agent-browser -p agentcore open https://x.com/home

# With explicit region
AGENTCORE_REGION=eu-west-1 agent-browser -p agentcore open https://example.com

# With SSO profile
AWS_PROFILE=my-sso-profile agent-browser -p agentcore open https://example.com
```

--------------------------------

### Profiler Commands: Start and Stop

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/profiling.md

Commands to initiate and terminate performance profiling. The start command can optionally include custom trace categories.

```bash
# Start profiling with default categories
agent-browser profiler start

# Start with custom trace categories
agent-browser profiler start --categories "devtools.timeline,v8.execute,blink.user_timing"

# Stop profiling and save to file
agent-browser profiler stop ./trace.json
```

--------------------------------

### Agent Browser Quick Start Commands

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Basic commands to open a URL, take a snapshot, interact with elements by reference, and close the browser.

```bash
agent-browser open example.com
agent-browser snapshot                    # Get accessibility tree with refs
agent-browser click @e2                   # Click by ref from snapshot
agent-browser fill @e3 "test@example.com" # Fill by ref
agent-browser get text @e1                # Get text by ref
agent-browser screenshot page.png
agent-browser close
```

--------------------------------

### Install Agent Browser Skills

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/skills/page.mdx

Use this command to install the discovery skill for agent-browser. This skill teaches your agent about agent-browser and directs it to use the `agent-browser skills` CLI command.

```bash
npx skills add vercel-labs/agent-browser
```

--------------------------------

### Start Session

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/dogfood/SKILL.md

Open the target URL in a named session and wait for the network to idle.

```bash
agent-browser --session {SESSION} open {TARGET_URL}
agent-browser --session {SESSION} wait --load networkidle
```

--------------------------------

### Install Lightpanda on Linux (x86_64)

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/engines/lightpanda/page.mdx

Installs the Lightpanda binary for Linux (x86_64 architecture). Make sure to make the downloaded binary executable and add it to your PATH.

```bash
curl -L -o lightpanda https://github.com/lightpanda-io/browser/releases/download/nightly/lightpanda-x86_64-linux && chmod a+x ./lightpanda
```

--------------------------------

### Plugin Manifest Response Example

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/plugins/page.mdx

Example response when retrieving plugin information using a manifest. It includes the plugin's name, capabilities, and description.

```json
{
  "protocol": "agent-browser.plugin.v1",
  "success": true,
  "manifest": {
    "name": "captcha",
    "capabilities": ["command.run", "captcha.solve"],
    "description": "Solve CAPTCHA challenges through Example CAPTCHA"
  }
}
```

--------------------------------

### Start Video Recording

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/changelog/page.mdx

Initiates video recording of browser sessions to a WebM file using Playwright.

```bash
agent-browser record start ./demo.webm
agent-browser click @e1
agent-browser record stop
```

--------------------------------

### List and Get Agent Browser Skills

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Commands to list available skills, retrieve their content, and find their directory paths. Use `--full` to include references and templates when getting skill content.

```bash
agent-browser skills                  # List available skills
agent-browser skills list             # Same as above
agent-browser skills get <name>       # Output a skill's full content
agent-browser skills get <name> --full  # Include references and templates
agent-browser skills get --all        # Output every skill
agent-browser skills path [name]      # Print skill directory path
```

--------------------------------

### Build Agent Browser from Source

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Builds the agent-browser CLI from source. Requires Node.js 24+, pnpm 11+, and Rust. Includes steps for cloning, installing dependencies, building, linking globally, and installing Chrome.

```bash
git clone https://github.com/vercel-labs/agent-browser
cd agent-browser
pnpm install
pnpm build
pnpm build:native   # Requires Rust (https://rustup.rs)
pnpm link --global  # Makes agent-browser available globally
agent-browser install
```

--------------------------------

### Initialize Directories

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/dogfood/SKILL.md

Create the necessary directory structure for storing screenshots and videos.

```bash
mkdir -p {OUTPUT_DIR}/screenshots {OUTPUT_DIR}/videos
```

--------------------------------

### Batching Agent Browser Commands

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/dogfood/SKILL.md

Batch independent agent-browser commands in a single shell call for efficiency, for example, to take a screenshot and then get the console output.

```bash
agent-browser ... screenshot ... && agent-browser ... console
```

--------------------------------

### Create Sandbox Snapshot

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/next/page.mdx

Run the helper script to generate a pre-configured VM image for faster startup.

```bash
npx tsx scripts/create-snapshot.ts
```

--------------------------------

### Navigate and Get Snapshot with agent-browser

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/page.mdx

Open a website, capture a snapshot of the accessibility tree, and interact with elements using their refs.

```bash
# Navigate and get snapshot
agent-browser open example.com
agent-browser snapshot -i

# Output:
# - heading "Example Domain" [ref=e1]
# - link "More information..." [ref=e2]

# Interact using refs
agent-browser click @e2
agent-browser screenshot page.png
agent-browser close
```

--------------------------------

### Run Agent Browser Doctor

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/installation/page.mdx

Use the doctor command to diagnose your agent-browser installation. Options include running a full diagnosis, a quick local-only check, performing repairs, or getting structured JSON output.

```bash
agent-browser doctor
agent-browser doctor --offline --quick
agent-browser doctor --fix
agent-browser doctor --json
```

--------------------------------

### Batch Pre-Navigation Setup Commands

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/init-scripts/page.mdx

Execute a sequence of commands, including opening a blank page, configuring network routes, setting cookies, and navigating to a target URL, before the main page loads.

```bash
agent-browser batch \
  '["open"]' \
  '["network","route","*","--abort","--resource-type","script"]' \
  '["cookies","set","--curl","cookies.curl","--domain","localhost"]' \
  '["navigate","http://localhost:3000/target"]'
```

--------------------------------

### AgentCore Session Output

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/providers/agentcore/page.mdx

Example of the Live View URL printed to stderr upon session initialization.

```text
Session: abc123-def456
Live View: https://us-east-1.console.aws.amazon.com/bedrock-agentcore/browser/aws.browser.v1/session/abc123-def456#
```

--------------------------------

### Verify Appium Installation

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/ios/page.mdx

Check the installed Appium version to confirm the installation was successful.

```bash
appium --version
```

--------------------------------

### Initialize Kernel session via CLI

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/providers/kernel/page.mdx

Use the -p flag to specify the provider and open a URL directly.

```bash
export KERNEL_API_KEY="your-api-key"
agent-browser -p kernel open https://example.com
```

--------------------------------

### Example Domain Allowlist with CDNs

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/security/page.mdx

Include CDN and third-party resource domains in your allowlist to prevent pages from breaking due to blocked sub-resource requests.

```bash
--allowed-domains "myapp.com,*.myapp.com,cdn.jsdelivr.net,fonts.googleapis.com,fonts.gstatic.com"
```

--------------------------------

### List and Show Configured Plugins

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/plugins/page.mdx

Verify that a plugin has been successfully added and is configured by listing all plugins or showing the details of a specific plugin.

```bash
agent-browser plugin list
agent-browser plugin show captcha
```

--------------------------------

### Start Agent Browser MCP Server

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/SKILL.md

Starts the Model Context Protocol (MCP) server for Agent Browser. Supports starting with all tools or specific tool profiles.

```bash
agent-browser mcp
agent-browser mcp --tools all
agent-browser mcp --tools core,network,react
```

--------------------------------

### React Commands Examples

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/react/page.mdx

Demonstrates common agent-browser commands for interacting with React applications, including inspecting the component tree, recording renders, and analyzing Suspense boundaries.

```bash
agent-browser react tree
agent-browser react inspect 42

agent-browser react renders start
agent-browser click @e3
agent-browser react renders stop

agent-browser react suspense --only-dynamic
```

--------------------------------

### Agent Browser Recording Commands

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/video-recording.md

Manage video recording sessions using start, stop, and restart commands. Ensure a session is launched before recording.

```bash
# Launch a session first
agent-browser open

# Start recording to file
agent-browser record start ./output.webm

# Stop current recording
agent-browser record stop

# Restart with new file (stops current + starts new)
agent-browser record restart ./take2.webm
```

--------------------------------

### Project Installation (Local Dependency)

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/installation/page.mdx

Installs Agent Browser as a local dependency for a specific project, allowing version pinning in package.json. Run 'agent-browser install' to download Chrome on the first use.

```bash
npm install agent-browser
npx agent-browser install  # Download Chrome (first time)
```

--------------------------------

### Cargo Installation (Rust)

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/installation/page.mdx

Installs Agent Browser using Cargo, compiling from source. Requires Node.js 24+, pnpm 11+, and the Rust toolchain. Run 'agent-browser install' to download Chrome on the first use.

```bash
cargo install agent-browser
agent-browser install  # Download Chrome (first time)
```

--------------------------------

### Create a Sandbox Snapshot for Fast Startup

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/vercel-sandbox/SKILL.md

This script creates a sandbox snapshot that includes system dependencies, agent-browser, and Chromium. This allows for sub-second sandbox startup times, significantly reducing execution overhead.

```typescript
import { Sandbox } from "@vercel/sandbox";

const CHROMIUM_SYSTEM_DEPS = [
  "nss", "nspr", "libxkbcommon", "atk", "at-spi2-atk", "at-spi2-core",
  "libXcomposite", "libXdamage", "libXrandr", "libXfixes", "libXcursor",
  "libXi", "libXtst", "libXScrnSaver", "libXext", "mesa-libgbm", "libdrm",
  "mesa-libGL", "mesa-libEGL", "cups-libs", "alsa-lib", "pango", "cairo",
  "gtk3", "dbus-libs",
];

async function createSnapshot(): Promise<string> {
  const sandbox = await Sandbox.create({
    runtime: "node24",
    timeout: 300_000,
  });

  await sandbox.runCommand("sh", [
    "-c",
    `sudo dnf clean all 2>&1 && sudo dnf install -y --skip-broken ${CHROMIUM_SYSTEM_DEPS.join(" ")} 2>&1 && sudo ldconfig 2>&1`,
  ]);
  await sandbox.runCommand("npm", ["install", "-g", "agent-browser"]);
  await sandbox.runCommand("npx", ["agent-browser", "install"]);

  const snapshot = await sandbox.snapshot();
  return snapshot.snapshotId;
}
```

--------------------------------

### Install Linux Dependencies

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/installation/page.mdx

Installs system dependencies required for Agent Browser on Linux systems.

```bash
agent-browser install --with-deps
```

--------------------------------

### Run agent-browser without installation

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/page.mdx

Execute agent-browser commands directly using npx without a global installation.

```bash
npx agent-browser open example.com
```

--------------------------------

### Setup Browserless with API Key

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/providers/browserless/page.mdx

Use this command to set your Browserless API key and open a URL using the Browserless provider. The -p flag explicitly specifies the provider.

```bash
export BROWSERLESS_API_KEY="your-api-token"
agent-browser -p browserless open https://example.com
```

--------------------------------

### Configure Browser Launch Arguments

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/changelog/page.mdx

Control browser startup with fine-grained options using the `--args`, `--user-agent`, and `--proxy-bypass` flags.

```bash
agent-browser --args "--disable-gpu,--no-sandbox" open example.com
agent-browser --user-agent "Custom UA" open example.com
agent-browser --proxy-bypass "localhost,*.internal" open example.com
```

--------------------------------

### Connect and Open Slack

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/slack/SKILL.md

Initialize a session by connecting to an existing browser instance or opening a new URL.

```bash
# Connect to existing session on port 9222 (typical for already-open Slack)
agent-browser connect 9222

# Or open Slack if not already running
agent-browser open https://app.slack.com
```

--------------------------------

### Open Channel for Review

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/slack/references/slack-tasks.md

Navigates to a channel and captures the initial state.

```bash
agent-browser click @e_channel_ref
agent-browser wait 1000
agent-browser snapshot -i
```

--------------------------------

### Optimal AI Workflow: Navigate, Snapshot, and Interact

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

A multi-step workflow demonstrating how to open a website, take an AI-parsable snapshot, and then interact with elements identified by references.

```bash
# 1. Navigate and get snapshot
agent-browser open example.com
agent-browser snapshot -i --json   # AI parses tree and refs

# 2. AI identifies target refs from snapshot
# 3. Execute actions using refs
agent-browser click @e2
agent-browser fill @e3 "input text"

# 4. Get new snapshot if page changed
agent-browser snapshot -i --json
```

--------------------------------

### Start HAR Export

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/network/page.mdx

Begin capturing network traffic for HAR export. Ensure routes are set before starting HAR export if needed.

```bash
agent-browser network har start
```

--------------------------------

### Global Installation of Agent Browser

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Installs the agent-browser CLI globally using npm. Requires a one-time command to download the necessary Chrome browser.

```bash
npm install -g agent-browser
agent-browser install  # Download Chrome from Chrome for Testing (first time only)
```

--------------------------------

### Launch Browser

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/commands.md

Launches the browser to about:blank. Use this command before other commands like network routing or setting cookies to prepare the browser's initial state.

```bash
agent-browser open
```

--------------------------------

### Generating Documentation with Video

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/video-recording.md

Create video recordings of user workflows for documentation purposes. Includes pauses to make the recorded actions easier to follow.

```bash
#!/bin/bash
# Record workflow for documentation

agent-browser open https://app.example.com/login
agent-browser record start ./docs/how-to-login.webm
agent-browser wait 1000  # Pause for visibility

agent-browser snapshot -i
agent-browser fill @e1 "demo@example.com"
agent-browser wait 500

agent-browser fill @e2 "password"
agent-browser wait 500

agent-browser click @e3
agent-browser wait --load networkidle
agent-browser wait 1000  # Show result

agent-browser record stop
```

--------------------------------

### Examples of Finding Elements with Actions

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/commands/page.mdx

Demonstrates practical usage of the `find` command with various semantic locators and actions. Options like `--name` and `--exact` can refine the search criteria.

```bash
agent-browser find role button click --name "Submit"
agent-browser find label "Email" fill "test@test.com"
agent-browser find alt "Logo" click
agent-browser find first ".item" click
agent-browser find last ".item" text
agent-browser find nth 2 ".card" hover
```

--------------------------------

### Core Browser Automation Workflow

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/quick-start/page.mdx

This sequence demonstrates the fundamental steps: opening a URL, taking a snapshot to identify elements, and interacting with those elements using their references. Re-snapshotting is shown after an interaction.

```bash
agent-browser open example.com
```

```bash
agent-browser snapshot -i
# Output:
# @e1 [heading] "Example Domain"
# @e2 [link] "More information..."
```

```bash
agent-browser click @e2
```

```bash
agent-browser snapshot -i
```

--------------------------------

### Start and Stop Agent Browser Dashboard

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Commands to start the dashboard server on the default or a custom port, and to stop it. The dashboard runs as a background process.

```bash
# Start the dashboard server (runs in background on port 4848)
agent-browser dashboard start
agent-browser dashboard start --port 8080   # Custom port

# All sessions are automatically visible in the dashboard
agent-browser open example.com

# Stop the dashboard
agent-browser dashboard stop
```

--------------------------------

### Inspect and Navigate an App

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/electron/SKILL.md

Pattern for launching, connecting, and navigating through an application's UI.

```bash
open -a "Slack" --args --remote-debugging-port=9222
sleep 3  # Wait for app to start
agent-browser connect 9222
agent-browser snapshot -i
# Read the snapshot output to identify UI elements
agent-browser click @e10  # Navigate to a section
agent-browser snapshot -i  # Re-snapshot after navigation
```

--------------------------------

### Video Recording Commands

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/commands.md

Record browser interactions as a video. This involves starting a browser session, starting the recording, performing actions, and then stopping the recording.

```bash
agent-browser open https://example.com     # Launch a browser session first
agent-browser record start ./demo.webm    # Start recording
agent-browser click @e1                   # Perform actions
agent-browser record stop                 # Stop and save video
agent-browser record restart ./take2.webm # Stop current + start new
```

--------------------------------

### Start Windows Debugging Instance

Source: https://github.com/vercel-labs/agent-browser/blob/main/AGENTS.md

Start the Windows Server 2022 EC2 instance if it is stopped. This command is used to make the debugging environment available.

```bash
./scripts/windows-debug/start.sh
```

--------------------------------

### Login using Credential Plugin

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/plugins/page.mdx

This bash command demonstrates how to log in to an application using a specified credential provider and item reference.

```bash
agent-browser auth login my-app --credential-provider vault --item "My App"
```

--------------------------------

### Cargo Installation of Agent Browser

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Installs the agent-browser CLI using Cargo, the Rust package manager. Requires a command to download the necessary Chrome browser.

```bash
cargo install agent-browser
agent-browser install  # Download Chrome from Chrome for Testing (first time only)
```

--------------------------------

### Start Recording Session

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/dogfood/SKILL.md

Initiates a recording session for interactive issue reproduction. Replace {SESSION} with a unique session identifier and {OUTPUT_DIR} with the desired output directory.

```bash
agent-browser --session {SESSION} record start {OUTPUT_DIR}/videos/issue-{NNN}-repro.webm
```

--------------------------------

### Install Lightpanda on macOS (Apple Silicon)

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/engines/lightpanda/page.mdx

Installs the Lightpanda binary for macOS with Apple Silicon architecture. Ensure the binary is executable and placed in your system's PATH.

```bash
curl -L -o lightpanda https://github.com/lightpanda-io/browser/releases/download/nightly/lightpanda-aarch64-macos && chmod a+x ./lightpanda
```

--------------------------------

### Execute Basic AgentCore Workflow

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/agentcore/SKILL.md

Standard commands for opening and interacting with a cloud-hosted browser session.

```bash
# Open a page on an AgentCore cloud browser
agent-browser -p agentcore open https://example.com

# Everything else is the same as local Chrome
agent-browser snapshot -i
agent-browser click @e1
agent-browser screenshot page.png
agent-browser close
```

--------------------------------

### Project Installation of Agent Browser

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Installs agent-browser as a local dependency for a project, allowing version pinning in package.json. Subsequent use can be via package.json scripts or direct invocation.

```bash
npm install agent-browser
agent-browser install
```

--------------------------------

### Launch and Navigate

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/commands.md

Launches the browser and navigates to a specified URL. Supports various protocols and automatically prepends 'https://' if no protocol is provided.

```bash
agent-browser open <url>
```

--------------------------------

### Open URL with Multiple Init Scripts

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/init-scripts/page.mdx

Specify multiple `--init-script` flags to run several initialization scripts in sequence before the page loads.

```bash
agent-browser --init-script ./a.js --init-script ./b.js open https://example.com
```

--------------------------------

### Get Snapshot with JSON Output

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Use the `--json` flag to get machine-readable output for the snapshot command. This is useful for parsing the browser's state and element references.

```bash
agent-browser snapshot --json
# Returns: {"success":true,"data":{"snapshot":"...","refs":{"e1":{"role":"heading","name":"Title"},...}}}
```

--------------------------------

### Configure Browserbase via CLI flag

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/providers/browserbase/page.mdx

Use the -p flag to specify the provider and set the API key directly in the command line.

```bash
export BROWSERBASE_API_KEY="your-api-key"
agent-browser -p browserbase open https://example.com
```

--------------------------------

### Implement Server Action for Browser Automation

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/next/page.mdx

A server action that manages the sandbox lifecycle, installs dependencies, and executes browser commands. Use a snapshot ID to avoid repeated installation of system dependencies.

```ts
"use server";
import { Sandbox } from "@vercel/sandbox";

const snapshotId = process.env.AGENT_BROWSER_SNAPSHOT_ID;

const CHROMIUM_SYSTEM_DEPS = [
  "nss", "nspr", "libxkbcommon", "atk", "at-spi2-atk", "at-spi2-core",
  "libXcomposite", "libXdamage", "libXrandr", "libXfixes", "libXcursor",
  "libXi", "libXtst", "libXScrnSaver", "libXext", "mesa-libgbm", "libdrm",
  "mesa-libGL", "mesa-libEGL", "cups-libs", "alsa-lib", "pango", "cairo",
  "gtk3", "dbus-libs",
];

function getSandboxCredentials() {
  if (
    process.env.VERCEL_TOKEN &&
    process.env.VERCEL_TEAM_ID &&
    process.env.VERCEL_PROJECT_ID
  ) {
    return {
      token: process.env.VERCEL_TOKEN,
      teamId: process.env.VERCEL_TEAM_ID,
      projectId: process.env.VERCEL_PROJECT_ID,
    };
  }
  return {};
}

async function withBrowser<T>(
  fn: (sandbox: InstanceType<typeof Sandbox>) => Promise<T>,
): Promise<T> {
  const credentials = getSandboxCredentials();

  const sandbox = snapshotId
    ? await Sandbox.create({
        ...credentials,
        source: { type: "snapshot", snapshotId },
        timeout: 120_000,
      })
    : await Sandbox.create({ ...credentials, runtime: "node24", timeout: 120_000 });

  if (!snapshotId) {
    await sandbox.runCommand("sh", [
      "-c",
      `sudo dnf clean all 2>&1 && sudo dnf install -y --skip-broken ${CHROMIUM_SYSTEM_DEPS.join(" ")} 2>&1 && sudo ldconfig 2>&1`,
    ]);
    await sandbox.runCommand("npm", ["install", "-g", "agent-browser"]);
    await sandbox.runCommand("npx", ["agent-browser", "install"]);
  }

  try {
    return await fn(sandbox);
  } finally {
    await sandbox.stop();
  }
}

export async function screenshotUrl(url: string) {
  return withBrowser(async (sandbox) => {
    await sandbox.runCommand("agent-browser", ["open", url]);

    const ssResult = await sandbox.runCommand("agent-browser", [
      "screenshot", "--json",
    ]);
    const ssPath = JSON.parse(await ssResult.stdout())?.data?.path;
    const b64Result = await sandbox.runCommand("base64", ["-w", "0", ssPath]);
    const screenshot = (await b64Result.stdout()).trim();

    await sandbox.runCommand("agent-browser", ["close"]);
    return { ok: true, screenshot };
  });
}

export async function snapshotUrl(url: string) {
  return withBrowser(async (sandbox) => {
    await sandbox.runCommand("agent-browser", ["open", url]);

    const result = await sandbox.runCommand("agent-browser", [
      "snapshot", "-i", "-c",
    ]);
    const snapshot = await result.stdout();

    await sandbox.runCommand("agent-browser", ["close"]);
    return { ok: true, snapshot };
  });
}
```

--------------------------------

### Login with Credential Provider Plugin

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Use a credential provider plugin for authentication. This example shows a basic login and a more detailed login with specific selectors and URL.

```bash
agent-browser auth login my-app --credential-provider vault --item "My App"
agent-browser auth login my-app --credential-provider vault --item "My App" --url https://app.example.com/login --username-selector "#email" --password-selector "#password" --submit-selector "button[type=submit]"
```

--------------------------------

### Capture Chrome DevTools Traces

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/debugging/page.mdx

Start a Chrome DevTools trace with `trace start`. End tracing and save the trace data to a JSON file using `trace stop [path]`. If no path is specified, a file is saved in the temp directory.

```bash
agent-browser trace start

agent-browser open https://example.com
agent-browser click @e3

agent-browser trace stop ./trace.json
```

--------------------------------

### Chrome Trace Event Format Example

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/profiler/page.mdx

This is an example of the JSON output generated by the profiler, conforming to the Chrome Trace Event format. It includes trace events with details like category, name, timestamp, duration, process ID, and thread ID.

```json
{
  "traceEvents": [
    {
      "cat": "devtools.timeline",
      "name": "RunTask",
      "ph": "X",
      "ts": 12345,
      "dur": 100,
      "pid": 1,
      "tid": 1
    }
  ],
  "metadata": {
    "clock-domain": "LINUX_CLOCK_MONOTONIC"
  }
}
```

--------------------------------

### Run Snapshot Creation Script

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/vercel-sandbox/SKILL.md

Execute the helper script to create a sandbox snapshot. This command simplifies the process of setting up the environment for fast startup.

```bash
npx tsx examples/environments/scripts/create-snapshot.ts
```

--------------------------------

### Get Text Content

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/slack/SKILL.md

Retrieve the text content of a specific element.

```bash
# Get a message or element's text
agent-browser get text @e_message_ref
```

--------------------------------

### Enable Native Rust Daemon

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/changelog/page.mdx

Demonstrates how to enable the experimental native Rust daemon for Agent Browser using command-line flags, environment variables, or configuration files.

```bash
# Via flag
agent-browser --native open example.com

# Via environment variable
export AGENT_BROWSER_NATIVE=1
agent-browser open example.com
```

--------------------------------

### Validate session names

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/sessions/page.mdx

Examples of valid and invalid session name formats.

```bash
# Valid session names
agent-browser --session-name my-project open example.com
agent-browser --session-name test_session_v2 open example.com

# Invalid (will be rejected)
agent-browser --session-name "../bad" open example.com    # path traversal
agent-browser --session-name "my session" open example.com # spaces
agent-browser --session-name "foo/bar" open example.com    # slashes
```

--------------------------------

### Install Vercel Sandbox dependency

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/next/page.mdx

Add the required Vercel Sandbox package to your project.

```bash
pnpm add @vercel/sandbox
```

--------------------------------

### Color Scheme and Download Options

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/commands/page.mdx

Options for setting the color scheme and default download directory. Use '--color-scheme' to specify 'dark', 'light', or 'no-preference'.

```bash
--color-scheme <scheme>  # Color scheme: dark, light, no-preference
--download-path <path>   # Default download directory
```

--------------------------------

### Snapshot ID Environment Variable

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/next/page.mdx

Example format for the snapshot ID environment variable.

```text
AGENT_BROWSER_SNAPSHOT_ID=snap_xxxxxxxxxxxx
```

--------------------------------

### Run Context Footprint Measurement

Source: https://github.com/vercel-labs/agent-browser/blob/main/evals/README.md

Execute the deterministic CLI vs MCP context footprint measurement.

```bash
bun run context-footprint.ts
```

--------------------------------

### Get Element Styles

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/changelog/page.mdx

Retrieves the computed styles of elements matching a given selector.

```bash
agent-browser get styles "button"
```

--------------------------------

### Connect to Cloud Browser Provider

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/cdp-mode/page.mdx

Use the -p flag to connect to a cloud browser provider instead of launching a local browser. This example connects to Browserbase.

```bash
agent-browser -p browserbase open https://example.com
```

--------------------------------

### Restart Video Recording

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/commands/page.mdx

Command to stop the current video recording and start a new one.

```bash
agent-browser record restart <path>   # Stop current and start new recording
```

--------------------------------

### Result Event Structure

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/dashboard/page.mdx

Example JSON structure for a result event sent when a command finishes.

```json
{
  "type": "result",
  "id": "r123",
  "action": "click",
  "success": true,
  "data": {},
  "duration_ms": 45,
  "timestamp": 1711367000045
}
```

--------------------------------

### Set Default Download Path

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/changelog/page.mdx

Opens a URL with a specified download path. Downloads will be saved to './downloads'.

```bash
agent-browser --download-path ./downloads open https://example.com
```

--------------------------------

### Start and Stop Tracing

Source: https://github.com/vercel-labs/agent-browser/blob/main/README.md

Use these commands to begin and end the recording of execution traces for debugging purposes.

```bash
agent-browser trace start             # Start recording trace
agent-browser trace stop [path]       # Stop and save trace
```

--------------------------------

### Workflow with Labeled Tabs and Actions

Source: https://github.com/vercel-labs/agent-browser/blob/main/skill-data/core/references/commands.md

Demonstrates a workflow using user-assigned labels for tabs, including creating labeled tabs, switching between them, and performing actions like clicking elements.

```bash
agent-browser tab new --label docs https://docs.example.com
agent-browser tab new --label app  https://app.example.com
agent-browser tab docs                   # switch to docs
agent-browser snapshot                   # populate refs for docs
agent-browser click @e1                  # ref click on docs
agent-browser tab app                    # switch to app
agent-browser tab close docs             # close by label
```

--------------------------------

### List Available iOS Simulators

Source: https://github.com/vercel-labs/agent-browser/blob/main/docs/src/app/ios/page.mdx

Use this command to see all iOS simulators available on your system, including their names and iOS versions.

```bash
agent-browser device list
```