This project provides a lightweight API enhancement service to integrate real-time web search capabilities into various Large Language Model (LLM) APIs, aiming to improve the accuracy and timeliness of AI responses. It is designed for users of LLM APIs who need to access up-to-date information.

How It Works

The service acts as a proxy, automatically detecting queries that require external information and performing real-time web searches. It then injects the retrieved search results into the prompt sent to the LLM, enabling the model to generate more informed and current answers. This approach leverages existing LLM infrastructure while augmenting it with dynamic data retrieval.

Quick Start & Requirements

• Direct Use: Add a new model in clients like Chatbox with the API address https://search-for-llmapi.dawne.cn/with-search/v1.
• Prerequisites: An API Key for your LLM platform and the correct model identifier (e.g., platform@model_name).
• Self-Deployment: Clone the repository, install dependencies (pip install flask requests), and run python server.py. SSL setup is required for server deployment.
• Documentation: https://eqrwxrl391e.feishu.cn/wiki/TEwxweXOOi5I2lk60VocY9QOnob

Highlighted Details

• Supports real-time web search for LLM queries.
• Compatible with multiple LLM platforms including SiliconFlow, Baidu Wenxin, ByteDance Volcengine, Alibaba Cloud Baishan, and OpenAI-compatible APIs.
• Features a "Chain-of-Thought" display for DeepSeek models.
• Includes automatic cleanup of historical thought processes.

Maintenance & Community

• Open to issue submissions for feedback and problem reporting.

Licensing & Compatibility

• The README does not specify a license.

Limitations & Caveats

• Self-deployment is currently incomplete and does not support the search functionality.