ZCode Agent Kit

ZCode Agent Kit
English (original) · Deutsch · Español · 日本語 · 简体中文

Use your ZCode account with the coding assistant you already use.

*The kit runs locally; model requests go to ZCode.*

The kit starts a small proxy on your machine. Your coding assistant talks to that proxy, and the proxy forwards the requests to ZCode with your own login, so you keep your assistant and use your existing ZCode models and quota. There are two ways to connect an assistant: let the kit configure the ones it detects, or copy the base URL and API key it prints into any OpenAI- or Anthropic-compatible client.

Step 1 — Check what you need

- ZCode Desktop, signed in to your own account with available model quota.
- Node.js 24 or newer. Check with node --version in a new terminal. Node.js 20 and 22 are no longer supported.
- A coding assistant, installed separately. The kit connects it to ZCode.

Step 2 — Install the kit

Pick one of the two methods and stay with it for updates. The release installer runs setup for you; with npm you run setup yourself.
Before you run the installer: it downloads and runs a script, and setup changes the configuration of the assistants you agree to. Setup also sends one small model request, which uses a little quota. Every change is recorded so it can be rolled back. If your security policy requires it, read the installer first.

Windows — PowerShell, without administrator rights:

macOS / Linux:

Run it from any directory; no clone is needed. The installer installs Bun 1.4.2 for the kit if Bun is missing. On Windows, use PowerShell, not Git Bash or WSL.

Alternative — npm (Windows, macOS and Linux):

Needs Bun on your PATH (bun --version; tested with Bun 1.4.2).

What setup asks

Setup looks for installed assistants and asks which of them it may configure. On…