Cloudflare Agents SDK Skill

The Cloudflare Agents SDK enables you to build and deploy AI-powered agents that run on Cloudflare's global edge network. Agents can autonomously perform tasks, communicate with clients in real-time via WebSockets, call AI models, persist state, schedule tasks, run asynchronous workflows, browse the web, query data from databases, support human-in-the-loop interactions, and act as MCP (Model Context Protocol) servers or clients.

Core Value: Agents deploy on Cloudflare's Durable Objects - stateful micro-servers that can scale to tens of millions and run close to users for low-latency interactivity.

When to Use This Skill

This skill should be triggered when:

• Building AI agents that need persistent state and real-time communication
• Creating chatbots or conversational AI interfaces on Cloudflare Workers
• Implementing WebSocket-based real-time applications
• Deploying agents with embedded SQL databases (SQLite)
• Scheduling tasks with delays, dates, or cron expressions
• Building human-in-the-loop approval workflows for AI tool execution
• Creating or connecting to MCP (Model Context Protocol) servers
• Deploying stateful AI applications on edge infrastructure
• Building multi-user collaborative AI applications
• Integrating with Vercel AI SDK on Cloudflare Workers