Show HN: Cleverb.ee – open-source agent that writes a cited research report

Summary

CleverBee: Advanced AI Research Assistant Website • Documentation • GitHub CleverBee is a powerful Python-based research agent using Large Language Models (LLMs) like Claude and Gemini, Playwright for web browsing, and Chainlit for an interactive UI. It performs research by browsing the web, extracting content (HTML), cleaning it, and summarizing findings based on user research topics. ✨ Features 🌐 Interactive web UI via Chainlit via Chainlit 🔧 MCP Tool Support: Integrates external tools via the Model Context Protocol (MCP) Integrates external tools via the Model Context Protocol (MCP) 🧠 Multi-LLM Research: Uses distinct, configurable LLMs for different tasks: Primary LLM: Gemini 2.5 Pro for planning and final report generation Next Step LLM: Gemini 2.5 Flash for analyzing research progress and deciding next actions Summarizer LLM: Gemini 2.0 Flash for intermediate web content summarization Uses distinct, configurable LLMs for different tasks: 🌍 Automated Web Browsing: Utilizes Playwright for searching the web and extracting HTML content Utilizes Playwright for searching the web and extracting HTML content 📊 Content Processing: Cleans HTML to Markdown before summarization Cleans HTML to Markdown before summarization 📈 Integrated Token Tracking: Monitors token usage and estimates costs for LLM calls Monitors token usage and estimates costs for LLM calls ⚙️ Highly Configurable: Settings managed via config.yaml Settings managed via 🚀 Modular LLM Clients: Supports different providers (Gemini, Claude, Local GGUF via llama-cpp-python) Supports different providers (Gemini, Claude, Local GGUF via llama-cpp-python) 💾 LLM Caching: Employs NormalizingCache (SQLite-based) for improved performance and cost reduction 🖥️ System Compatibility macOS: Fully tested and supported, including both Intel and Apple Silicon (via Rosetta 2) Fully tested and supported, including both Intel and Apple Silicon (via Rosetta 2) Linux: Fully supported, with NVIDIA GPU detection and optimization for...