Auto-Coder.Chat Docs

Introduction

Auto-Coder.Chat — the world's first Async Vibe Coding agent, offering both CLI and local Web UI modes, ready to use instantly.

Introduction to Auto-Coder.Chat

Auto-Coder.Chat is the world's first Async Vibe Coding agent, built for developers who want AI-powered coding without limits. It offers two ways to work — CLI and local Web UI — with zero cloud deployment or complex setup required. Just pip install, navigate to your project directory, and start coding. Instant access, anytime.

Two Modes, Your Choice

CLI Mode — The Home of Async Vibe Coding

Run auto-coder.chat in your terminal to enter an interactive AI coding session. CLI mode is the core of Auto-Coder.Chat, purpose-built for:

  • Async Vibe Coding — Submit coding tasks to run in the background via /async. Multiple tasks execute in parallel, and results are automatically merged into the main branch — even when tasks conflict.
  • Long-Running Tasks — Precisely control AI runtime with the /time parameter (e.g., /async /name <job_name> /time 60m). Perfect for large-scale refactoring, batch code generation, and other scenarios that require extended execution.
  • Flexible SubAgent Workflow — Users or agents can define custom SubAgent workflows, breaking complex tasks into multiple sub-tasks executed in stages for efficient cost control and task orchestration.

Local Web UI — Friendlier Than CLI, Lighter Than IDE

Run auto-coder.web to open the web interface in your browser, ready to launch and use at any time:

  • Minimal Dev View — Chat with AI via the Web UI, monitoring task progress and results in real time.
  • Pro Dev View — Built-in code editor, file tree, and terminal for a full IDE experience.

Core Advantages

  • Async Vibe Coding — Run coding tasks in the background and automatically merge multiple conflicting tasks into the main branch. Free yourself from waiting.
  • Controllable AI Duration — Control AI runtime with the /time parameter to avoid wasting resources.
  • Unlimited Context — Run sessions over 800k tokens using 128k/200k context models with intelligent chunking.
  • SubAgent Workflow — Define custom SubAgent workflows for flexible task orchestration and efficient cost control.
  • Domestic Model Support — Support for coding plan subscriptions with domestic models including DeepSeek, GLM5, M2.7, and more.
  • 100% Open Source — Fully open source, freely customizable, no restrictions.

Quick Start

# Install
pip install -U auto-coder auto_coder_web

# Navigate to your project
cd your-project

# Launch CLI mode
auto-coder.chat

# Or launch Web UI
auto-coder.web

Next Steps

Ready to get started? Head to the Quick Start guide and get up and running in minutes.

下一篇
Quick Start