Flask Webapplicatie met LLM-integratie en MCP-tools

What is Flask Webapplicatie met LLM-integratie en MCP-tools?

Flask Webapplicatie met LLM-integratie en MCP-tools is a local Flask web application that provides a browser-based interface to access OpenAI GPT-3.5 and Anthropic Claude 2 LLMs. It integrates MCP (Model Context Protocol) servers for Brave Search and GitHub, automatically enriching user prompts with context from these tools before sending them to the selected model. Built with Python 3.7+, it runs on Windows, macOS, and Linux and is intended for developers who want to combine LLMs with live external data.

How to use Flask Webapplicatie met LLM-integratie en MCP-tools?

1. Clone the repository, create a Python virtual environment, and install dependencies via pip install -r requirements.txt.
2. Configure API keys either in a .env file (OPENAI_API_KEY, ANTHROPIC_API_KEY, BRAVE_API_KEY, optional GITHUB_TOKEN) or as environment variables.
3. Run the application with python app.py – it will start on http://localhost:5000.
4. In the web interface, optionally start the Brave Search and/or GitHub MCP servers, select an LLM model, type a prompt, and click “Verstuur naar AI” to get an enriched answer. You can also view the full prompt with added context.

Key features of Flask Webapplicatie met LLM-integratie en MCP-tools

• Choose between OpenAI GPT-3.5 and Anthropic Claude 2 models
• Input prompts and receive LLM responses directly in the browser
• Start and stop external MCP tools (Brave Search, GitHub) from the web interface
• Automatically enrich prompts with relevant context from active MCP servers
• View the complete prompt including the added context before the LLM answers

Use cases of Flask Webapplicatie met LLM-integratie en MCP-tools

• Retrieve live web search results via Brave Search and include them as context for a more informed LLM answer
• Look up GitHub repositories, issues, or pull requests and have the LLM generate summaries or insights based on that data
• Compare responses from GPT-3.5 and Claude 2 when both are enriched with the same external context
• Prototype or demonstrate how MCP servers can enhance LLM prompts without requiring a complex backend setup

FAQ from Flask Webapplicatie met LLM-integratie en MCP-tools

What is MCP and why is it used here?

MCP (Model Context Protocol) is an open protocol that allows AI models to securely interact with external resources via standardised server interfaces. In this app, MCP servers gather extra context (e.g., web search results or GitHub data) based on the user prompt and add it to the request before it reaches the LLM.

Which API keys are required to run the application?

You need an OpenAI API key for GPT-3.5, an Anthropic API key for Claude 2, and a Brave Search API key for the Brave Search MCP server. A GitHub personal access token is optional (used for higher API rate limits). All keys can be set via a .env file or as environment variables.

Does it work on Windows, macOS, and Linux?

Yes. The application is pure Python with standard Flask and requests dependencies, and the README provides shell and PowerShell commands for setting up virtual environments and environment variables on all three platforms.

Is this application ready for production?

No. The README explicitly notes that the MCP server implementations are simplified for demonstration purposes and recommends better error handling and security measures for production use.

What are the known limitations?

The MCP servers are simplifications, the app requires valid API keys for every external service, and there is no built-in user authentication or advanced error recovery. The application is intended for local prototyping and learning.