Hello legends — in this tutorial I walk you through Claude Managed Agents end-to-end, from the Claude console to a real deployment. I start by spinning up a brand new managed agent from scratch and explain the core building blocks you actually need to understand: agents, sessions, environments, and the credentials vault. The big “aha” here is sessions: you can run one customer support agent, and every new customer conversation becomes its own private session with its own context. Then I show a practical integration: an n8n web chat as the front end, with the Claude agent as the back end. We create a session, send a message, wait for the agent to do its thing (tool calls, MCP connections, files), and then list session events to grab the latest assistant reply. I also break down pricing in plain English: you’re paying for tokens plus runtime at $0.08 per session hour, and you’re only charged while the agent is actively running — not while it’s idle. By the end, you’ll know how to configure network access (like restricting to your domain), connect OAuth tools like ClickUp, test inside the console, and then stitch the same flow together in n8n with a simple session-ID mapping so conversations stay persistent.