Bing Flights MCP Server

A Model Context Protocol (MCP) server that scrapes flight information from Bing Flights using Playwright. This project provides both a standalone Python scraper module and an MCP server wrapper for integration with MCP-compatible applications.

📦

Features

• 🔍 Search for one-way and round-trip flights

• ✈️ Support for multiple passengers (adults, children, infants)

• 💺 All cabin classes (Economy, Premium Economy, Business, First)

• 🤖 Headless or headed browser modes

• 📊 Structured JSON responses

• 🔧 Easy integration via MCP protocol

Project Structure

Installation

Quick Start (Recommended)

The easiest way to use this MCP server is with uvx:

This will automatically install the package and its dependencies in an isolated environment.

Installation via pip

You can also install from PyPI:

After installation, install the Playwright browser:

Development Installation

For development or if you want to modify the code:

1. Clone or download this repository

2. Create and activate a virtual environment

   On Windows:

   python -m venv venv venv\Scripts\activate

   On macOS/Linux:

   python -m venv venv source venv/bin/activate

3. Install dependencies

   pip install -r requirements.txt

4. Install Playwright browsers

   playwright install chromium

Usage

As an MCP Server

Using uvx (Recommended)

If you installed via PyPI, run:

Or add to your MCP settings configuration:

Running from Source

If you're developing or running from source:

Available Tools

The MCP server exposes two tools:

1.

Search for flight options from Bing Flights.

Parameters:

• origin (string, required): Origin airport code (e.g., "SEA")

• destination (string, required): Destination airport code (e.g., "ICN")

• departure_date (string, required): Departure date in YYYY-MM-DD format

• return_date (string, optional): Return date for round-trip searches

• adults (integer, optional, default=1): Number of adult passengers

• children (integer, optional, default=0): Number of child passengers

• infants (integer, optional, default=0): Number of infant passengers

• cabin_class (integer, optional, default=0): 0=Economy, 1=Premium Economy, 2=Business, 3=First

• max_results (integer, optional, default=10): Maximum number of results

• headless (boolean, optional, default=true): Run browser in headless mode

Example MCP Tool Call:

2.

Check scraper health and configuration.

Example Response:

As a Standalone Python Module

You can also use the scraper directly in your Python code:

Response Format

The scraper returns results in the following JSON structure:

Configuration Options

Cabin Classes

• 0 - Economy

• 1 - Premium Economy

• 2 - Business

• 3 - First Class

Browser Modes

• headless=True - Browser runs in the background (faster, no UI)

• headless=False - Browser window visible (useful for debugging)

Troubleshooting

Common Issues

Issue: Playwright browser not found

Issue: Timeout waiting for flight results

Issue: No results returned

Issue: Import errors

Debugging

To debug scraping issues, run with headless=False:

This will show the browser window so you can see what's being loaded.

Running Tests

Basic Tests

Run the basic unit tests that verify module imports and URL construction:

End-to-End Tests

Run comprehensive end-to-end tests that perform actual flight searches:

Note: E2E tests make real web requests to Bing Flights and may take several minutes to complete. They include:

1. Invalid Parameters Test - Verifies error handling

2. One-Way Search Test - Real search from SEA to ICN

3. Round-Trip Search Test - Real search from LAX to JFK

4. Multiple Passengers Test - Search with 2 adults and 1 child

The tests use headless browser mode and future dates to ensure valid searches.

Technical Details

Web Scraping Approach

• Uses Playwright for browser automation

• Waits for dynamic content to load

• Extracts data from Bing Flights result cards

• Handles multiple selector patterns for robustness

Error Handling

The scraper uses minimal error handling and allows exceptions to propagate:

• Network errors → Raises exception

• Timeout errors → Raises PlaywrightTimeoutError

• Invalid parameters → Raises ValueError

• Parsing errors → Returns partial results or empty list

This design allows the MCP client to implement appropriate retry logic and error recovery.

Limitations

• Only returns outbound flight details (return flight info not available on Bing results page)

• Maximum results limited by what's visible on the initial page load

• No pagination support (first page results only)

• Scraping depends on Bing's page structure (may break if they change their HTML)

Dependencies

• fastmcp>=0.1.0 - MCP server framework

• playwright>=1.40.0 - Browser automation

• python-dateutil>=2.8.2 - Date parsing utilities

Contributing

When contributing, please:

1. Test changes with both headless and headed modes

2. Verify compatibility with the MCP protocol

3. Update documentation for new features

4. Follow existing code style and patterns

License

MIT License - See LICENSE file for details

Disclaimer

This tool scrapes publicly available data from Bing Flights. Please:

• Use responsibly and respect rate limits

• Review Bing's Terms of Service

• Do not use for commercial purposes without proper authorization

• Be aware that web scraping may break if the website changes

Support

For issues, questions, or contributions, please open an issue on the project repository.