Run the same task through the Agent SDK

The same charges-register task as the Claude Code trace, run through query() from the Claude Agent SDK with only Read and Glob, on any OpenRouter model or a local Ollama model. The loop takes query as a parameter, so a demo can replay canned messages instead of calling a model.

Back to the exercise

Before you start

Before you start:Install Claude Code Node.js, pnpm and TypeScript Get an OpenRouter key OpenRouter or a local model?

This exercise calls a model. Run it either way

Any model through OpenRouterOne key for Claude, GPT, Gemini, Llama and more. Pay per token, with a credit limit.A local open-source modelFree and private with Ollama. Weaker on hard tasks.

Which should I pick? →

query() from the Claude Agent SDK runs the same loop as Claude Code, so the agent is a prompt, a list of tools and a turn limit. Two options make the comparison with your Claude Code trace fair: tools: ["Read", "Glob"] limits what the model can see, and settingSources: [] stops the SDK loading any CLAUDE.md or settings from your machine. The SDK does not run on a Claude subscription, so the model settings come from a .env file that points at OpenRouter or at a local model through Ollama, with no code change between them.

Answers Lesson 1: What is an agent?

Step-by-step answer

Stretch 30 min6 steps