# Chromedp Mcp

> MCP server for browser automation using chromedp

- **Type:** MCP server
- **Install:** `agentstack add mcp-kepatrick-chromedp-mcp`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [kepatrick](https://agentstack.voostack.com/s/kepatrick)
- **Installs:** 0
- **Category:** [Web & Browser](https://agentstack.voostack.com/c/web-and-browser)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [kepatrick](https://github.com/kepatrick)
- **Source:** https://github.com/kepatrick/chromedp-mcp

## Install

```sh
agentstack add mcp-kepatrick-chromedp-mcp
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# chromedp-mcp

A Model Context Protocol (MCP) server that provides browser automation capabilities using chromedp. This server enables AI assistants to interact with web pages, manage browser instances, and perform various web automation tasks.

## ⚠️ Development Status

**This project is currently under development. Some features may be incomplete, unstable, or contain bugs. Use with caution and please report any issues you encounter.**


## Features

- **Multi-instance Chrome Management**:  Create and manage multiple Chrome browser instances with extensive configuration options (headless mode, security settings, performance optimization)
- **Complete Page Navigation**: Navigate to URLs, browser history navigation (back/forward)
- **Element Interaction**: Find, click, and interact with web page elements using various selectors
- **Form Operations**: Input text, set values directly
- **Cookie Management**: Set and manage browser cookies for session handling
- **File Operations**: Download files and images from web pages
- **PDF Generation**: Convert web pages or HTML content to PDF documents
- **Keyboard Simulation**: Send keyboard events, key combinations, and modifier keys with comprehensive key support
- **Screen Capture**: Take screenshots of web pages for visual analysis
- **DOM Tree Extraction**: Get clean DOM structures without scripts/styles for analysis

## LLM Decision Support

**chromedp-mcp** provides web page information to LLMs, enabling AI to make decisions:

### Visual Information
**Provided through `screenshot` tool:**
- **Complete page screenshots**: Actual visual representation of web pages including layout, colors, images
- **Element positioning**: LLMs can see exact positions and appearance of buttons, links, forms and other elements
- **Page state identification**: Detect loading states, popups, error messages and other visual indicators
- **Responsive layout understanding**: Comprehend page presentation across different viewport sizes

### Structured Information  
**Provided through DOM-related tools:**
- **Clean DOM structure** (`get-all-elements`): Pure HTML structure with scripts and styles removed
- **Specific element details** (`select-element`): Element tree structure at specified depth
- **Text content extraction** (`get-element-withtext`): Element information containing specific text
- **Element attribute data**: ID, class, data-* and other attribute values

## Available Tools

For detailed tool specifications,and parameters, see [Tools Specification](doc/tools-spec.md)

### Instance Management
- **`create-chrome-instance`** - Create a new Chrome browser instance with extensive configuration options (headless mode, security settings, performance options)
- **`close`** - Close a specific Chrome browser instance

### Page Navigation  
- **`navigate`** - Navigate to a specified URL and return clean DOM structure
- **`navigate-back`** - Navigate to previous page in browser history
- **`navigate-forward`** - Navigate to next page in browser history

### Element Operations
- **`get-element-withtext`** - Find and retrieve information about a specific element with text content
- **`get-all-elements`** - Get all elements of current page as clean DOM tree structure
- **`select-element`** - Select element by CSS selector and return clean DOM structure at specified depth
- **`click-element`** - Click on an element with support for different click types (left, right, double)

### Visual Information Access

- **`screenshot`** - Capture screenshots of web pages for visual analysis and layout understanding


### Input Operations
- **`send-key`** - Send keyboard input to specified elements
- **`set-value`** - Directly set the value of form elements
- **`key-event`** - Send specific keyboard events and combinations with modifier keys

### Cookie Management
- **`set-cookie`** - Set HTTP cookies with full control over domain, path, security, and expiration settings

### File Operations
- **`download-file`** - Download files by clicking download links or buttons
- **`download_image`** - Download images from URL or by selector

### Document Generation
- **`generate_pdf`** - Generate PDF from HTML content or URL

### Other
- **`tips`** - Important usage tips and best practices for Chrome automation tools (recommended to check first)



## Requirements

- **Chrome/Chromium Browser**: Must be installed and accessible in system PATH
- **Go**: Version 1.19 or higher for building from source

## Environment Configuration

The following environment variables can be configured for Chrome management:

| Variable | Description | Default |
|----------|-------------|---------|
| `CHROME_MAXIMUM_INSTANCE` | Maximum concurrent Chrome instances | `5` |
| `CHROME_TTL` | Instance time-to-live in minutes | `15` |
| `CHROME_EXE_TIMEOUT` | Operation timeout in seconds | `300` |

These can be configured through your MCP client configuration or environment variables.

## Setup

### 1. Clone the repository:
   ```bash
   git clone https://github.com/KePatrick/chromedp-mcp.git

   cd chromedp-mcp
   ```

### 2. Build the project:
   **Windows**
   ```bash
   go build -o chromedp-mcp.exe .\cmd\chromedp-mcp\main.go
   ```
   **Linux/MacOs**
   ```bash
   go build -o chromedp-mcp ./cmd/chromedp-mcp/main.go
   ```


### 3. MCP Client Configuration

Configure your MCP-compatible client to use this server:

**Basic Configuration:**
```json
{
  "mcpServers": {
    "chromedp-mcp": {
      "command": "/path/to/chromedp-mcp",
      "args": []
    }
  }
}
```

**With Environment Variables:**
```json
{
  "mcpServers": {
    "chromedp-mcp": {
      "command": "/path/to/chromedp-mcp",
      "args": [],
      "env": {
        "CHROME_MAXIMUM_INSTANCE": "10",
        "CHROME_TTL": "30",
        "CHROME_EXE_TIMEOUT": "600"
      }
    }
  }
}
```

## Important Notes

- Always close instances when done to free resources
- Respect website terms of service and robots.txt
- Some websites may have anti-automation measures

## Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [kepatrick](https://github.com/kepatrick)
- **Source:** [kepatrick/chromedp-mcp](https://github.com/kepatrick/chromedp-mcp)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/mcp-kepatrick-chromedp-mcp
- Seller: https://agentstack.voostack.com/s/kepatrick
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.