Simple Search Engine App Using Langchain + DuckDuckGo

This project is a minimalist search tool that integrates DuckDuckGo with Langchain to enable real-time LLM-powered information retrieval.

It leverages the DuckDuckGoSearchRun tool from langchain_community.tools and supports multiple LLMs using Azure OpenAI APIs.

💡 Core Langchain Concepts Used:

• Models -> The LLMs (like OpenAI,Gemini or Azure OpenAI) used to generate or process language-based output.
• Prompts -> Instructions or input templates provided to the model to guide its responses.
• Chains -> Sequences of calls (like prompts + models) that execute in order to perform a task.
• Indexes -> Data structures (like vector stores) that help in efficient document retrieval and search.
• Memory -> Enables the application to remember past interactions across conversations or requests.
• Agents -> Components that use tools and models to decide dynamically what actions to take based on input.

The architecture is modular, and the backend is built using Flask. Users can test search queries by running a local server and using curl or frontend tools.

🧪 Example usage: curl -X POST http://localhost:5000/news -H "Content-Type: application/json" -d '{"query": "Hiking tips to mountains"}'

The project uses environment variables to protect sensitive keys. Make sure to check out the included .env-sample and setup accordingly.

🔗 Source Code: GitHub Repository