GitHub - TrippingKelsea/weblookrs

WebLook

A command-line tool for capturing screenshots and recordings of web pages.

Why This Tool Exists

I created WebLook to support my local development workflow, particularly when testing Amazon Q CLI and MCP (Model Context Protocol) applications. I needed a lightweight, read-only web tool with sane defaults that could quickly capture visual states of locally-running web applications without complex configuration. WebLook is designed to be simple, efficient, and integrate seamlessly into development pipelines with minimal overhead.

Features

Take screenshots of web pages
Create animated GIF recordings of web pages
Configurable wait time before capture
Configurable window size
Configurable recording length
Support for input/output piping
Headless operation
Execute custom JavaScript before capture
Capture browser console logs
Automatic user-agent rotation (Windows/Mac Chrome)
Automatic ChromeDriver management
Colorful progress indicators with countdown timers
[EXPERIMENTAL] MCP (Model Context Protocol) integration for AI model interaction

Usage

Options

--output, -o <FILE>: Specify output file (default: weblook.png or weblook.gif)
--wait, -w <SECONDS>: Wait time before capture (default: 10 seconds)
--record, -r [SECONDS]: Create a recording instead of screenshot (default length: 10 seconds)
--size, -s <WIDTHxHEIGHT>: Set viewport size (default: 1280x720)
--js, -j <CODE>: Execute JavaScript code before capture
--console-log <FILE>: Capture browser console logs and save to specified file
--debug, -d: Enable debug output (shows ChromeDriver messages)
--mcp-server <HOST:PORT>: [EXPERIMENTAL] Start as MCP server on specified address
--mcp-client <URL>: [EXPERIMENTAL] Connect to MCP server at specified URL
--help, -h: Show help information

Examples

# Take a screenshot of the default URL (127.0.0.1:8080)
weblook

# Take a screenshot of a specific URL
weblook https://example.com

# Take a screenshot after waiting 5 seconds
weblook --wait 5 https://example.com

# Create a 5-second recording
weblook --record 5 https://example.com

# Set viewport size to 1920x1080
weblook --size 1920x1080 https://example.com

# Execute JavaScript before capture
weblook --js "document.body.style.backgroundColor = 'red';" https://example.com

# Capture console logs to a file
weblook --console-log console.log https://example.com

# Pipe URL input and output to another command
echo "https://example.com" | weblook --output - | other-command

# Save output to a specific file
weblook https://example.com --output screenshot.png

# Show debug output
weblook --debug https://example.com

# [EXPERIMENTAL] Start as an MCP server
weblook --mcp-server 127.0.0.1:8000

# [EXPERIMENTAL] Use as an MCP client
weblook --mcp-client http://localhost:8000 https://example.com

Installation

WebLook is currently not available on crates.io. To install:

# Clone the repository
git clone https://github.com/username/weblook.git
cd weblook

# Build and install without experimental MCP support
cargo build --release

# Build with experimental MCP support
cargo build --release --features mcp_experimental

# The binary will be available at target/release/weblook
# You can copy it to a directory in your PATH for easier access
cp target/release/weblook ~/.local/bin/  # or another directory in your PATH

Requirements

ChromeDriver must be installed
- Install ChromeDriver: sudo apt install chromium-chromedriver (Ubuntu/Debian)
- The application will automatically start and stop ChromeDriver as needed

Experimental Features

MCP (Model Context Protocol) Integration

The MCP integration is currently experimental and requires compiling with the mcp_experimental feature flag. This feature allows WebLook to:

Act as an MCP server that other applications can connect to
Act as an MCP client that can connect to other MCP servers

To enable MCP support:

cargo build --features mcp_experimental

License

GPL-3.0