Empowers AI agents to dynamically discover and invoke containerized tools, registered at runtime, leveraging Cloudflare's serverless infrastructure.
Sponsored
This proof-of-concept project demonstrates a dynamic Model Context Protocol (MCP) server built on Cloudflare Workers, Containers, and D1. It enables AI agents, such as Claude, to discover and invoke containerized tools without requiring redeployment of the core service. Tools are registered via a simple HTTP API and stored in D1, with each tool running in an isolated Cloudflare Container, offering secure isolation and serverless scale-to-zero capabilities. The architecture provides a flexible way to extend AI agent functionality with custom or external tools.
Key Features
01Containerized Execution in isolated Cloudflare Containers
02HTTP API for runtime tool registration
030 GitHub stars
04Dynamic Tool Registry via Cloudflare D1
05MCP Protocol for AI agent tool discovery
06Serverless Scale-to-Zero for efficient resource usage
Use Cases
01Extending AI agent capabilities with custom or external tools
02Building a dynamic and scalable tool registry for AI models
03Registering new tools for AI interaction without service redeployment