Building an AI Agent that puts humans in the loop with Knock and Cloudflare’s Agents SDK

What this post added

This post introduces the integration of Knock's Agent Toolkit with Cloudflare's Agents SDK to build AI agents that incorporate human-in-the-loop approval workflows. It details how to expose tools to an LLM, wrap them with Knock's `requireHumanInput` for deferring execution until human approval, and handle asynchronous responses via webhooks. The example demonstrates a virtual card issuing workflow requiring administrative approval.