Quick Start

This page focuses only on the steps most users need the first time they use YutoCode. It does not expand on internal implementation.

Understand What It Is First

YutoCode is positioned as: the next-generation AI coding agent that breaks platform lock-in and works natively with OpenAI, Claude, Gemini, and OpenClaw.

In practical terms, that means two things:

• your workflow does not need to be locked into one platform
• you can keep one terminal-native way of working across different models and ecosystems

That is the largest difference between YutoCode and ordinary single-platform AI coding tools.

Step 1: Choose One Provider

On first use, do not optimize for trying every compatible ecosystem at once. A more reliable approach is to choose one provider for which you already have credentials and can complete the login path quickly.

Common options:

• Yuto API: suitable for teams already using YutoAI capabilities
• OpenAI Codex: suitable for users who already have the corresponding account and authorization flow
• Anthropic: suitable for users who already have access to Claude or the Anthropic API

Getting one full path working is more important than connecting every platform on day one.

Step 2: Log In

Using Yuto API

If you are using an OpenAI-compatible interface:

yuto auth login --provider yuto --api-format openai --api-key YOUR_YUTO_API_KEY

If you are using a Claude-format interface:

yuto auth login --provider yuto --api-format claude --api-key YOUR_YUTO_API_KEY

Using OpenAI Codex

yuto auth login --provider openai-codex

Once the browser authorization finishes, paste the final redirected http://localhost:1455/auth/callback?... URL back into the terminal exactly as received.

Using Anthropic

yuto auth login --provider anthropic --claudeai

If you already have an API key, you can save it directly:

yuto auth login --provider anthropic --api-key YOUR_ANTHROPIC_API_KEY

Step 3: Verify Login Status

yuto auth status --json

If you use multiple accounts or environments, manage them through profiles:

yuto auth profiles --json
yuto auth use-profile work

Step 4: Start the First Session

Inside a project directory, run:

yuto

The first session is best spent on three simple tasks:

1. ask it to summarize the repository structure
2. ask it to explain a file or a section of code
3. ask it to propose one clearly scoped change

For example:

Inspect the directory structure of this repository first, then tell me which files are most likely to contain the login flow.

Step 5: If You Do Not Want a Persistent Session

Use one-shot mode directly:

yuto --print "Summarize the purpose of this repository."

This is useful for quick summaries, explanations, or first-pass analysis.

The Most Useful Slash Commands in REPL

• /help
• /status
• /provider
• /model
• /tools
• /config
• /clear
• /exit

You do not need to memorize everything on the first day. Start with status, provider, and exit.