AgentBox

agentbox

Your AI coding workspace. On your server. In your browser.

Run Claude Code and Codex CLI on a Linux server you control. Go from a task to reviewed code with chat, terminal, files, and Git in one place.

English | 简体中文

Watch the tour · Highlights · Screenshots · Quick start · Architecture · Documentation

See it work

▶ Full tour in 1920 px (MP4, 70 s) — adds file previews, usage records and switching between the six interface styles · Screenshots

Recorded from the real web console with synthetic projects, conversations, command output and usage — no model was called. It demonstrates the interface, not model quality or speed.

Why agentbox?

Set up your coding environment once on your server, then pick up your project from a laptop, desktop, or phone. Your local machine only needs a browser; the agents and project tools run in Docker on the server.

You bring the server and your own Claude / Codex subscription, API key, or compatible relay account. Agentbox provides the workspace and management layer; model usage follows your provider's terms and pricing.

Highlights

A chat that keeps up with the CLI

- Watch the work, not a spinner. Replies stream with thinking summaries and every tool call; Claude Code tool calls read like its terminal (● Update(src/app.ts) with a numbered diff, reads and commands folded into "Read 3 files, ran 2 shell commands", sub-agents with their own calls), Codex commands are named the way its CLI shows them (Read, Search, Shell, Edit) with exit status and duration, and full output and diffs are a click away. The agent's task list sits above the message box ("Tasks 2/3 · Running the tests"), a timer counts the turn, and a ring shows how full the context window is, like the CLI status line.
- @ files and / commands. Type @ to search…