A serverless, AI-powered deep research agent built with Cloudflare Workers and Google Gemini 2.5

workers-research

A serverless, Cloudflare Workers-based Deep Research Agent powered by Google Gemini 2.5.

✨ Features

• In-depth Research with Google Gemini 2.5: Leverages Google's latest models for high-quality research reports.
• Serverless on Cloudflare: Built with Workers and Workflows for scalability, reliability, and cost-effectiveness.
• Intuitive Dashboard: Manage and view research via a Hono/JSX web interface.
• Simplified Architecture: Gemini 2.5's large context window removes the need for prompt compression, enhancing report quality.
• AutoRAG Integration: Allows the system to perform research using Cloudflare AI's AutoRAG feature, which queries against a pre-indexed dataset. This can be used as an alternative or in conjunction with standard web browsing research. The AutoRAG feature is invoked when an autorag_id is provided with the research request.
• Inline Q&A: Ask follow-up questions about completed reports, with answers grounded in the report content and saved for future reference.

🚀 Inspiration and Acknowledgements

Inspired by dzhng's deep-research. Thank you to dzhng for the original concept. workers-research adapts this for the Cloudflare ecosystem.

💡 Key Differences from Original Deep-Research

• Cloudflare Native: Runs on Cloudflare Workers and Workflows, unlike the Node.js/Docker original.
• Enhanced Reliability: Uses Cloudflare Workflows for robust multi-step research.
• Simplified & Efficient: Leverages Gemini 2.5's large context window, omitting prompt compression for a streamlined design and potentially better reports.
• Accessible Crawling: Utilizes Cloudflare's Browser Rendering (available on the free Workers plan).

🔍 Use Cases

• Academic Research: Quickly gather i