CLI Interactive Chat

CLI Interactive Chat

Quick Reference

• Who: Developer
• Where: Terminal → codebuff command
• Time: ~2 minutes to start
• Prerequisites: npm, Node.js 18+ or Bun

Prerequisites

• Installed Codebuff CLI (npm install -g codebuff)
• Have a project directory to work in
• Internet connection (for LLM API calls)

Step-by-Step Guide

Step 1: Install & Launch

1. Install Codebuff globally:

   npm install -g codebuff

2. Navigate to your project:

   cd your-project

3. Launch Codebuff:

   codebuff

Tip

First time? You’ll be prompted to log in via GitHub OAuth. A browser window will open automatically.

Step 2: Chat with the Agent

1. Type your request in natural language:

   Add error handling to all API endpoints

2. Watch as Codebuff:

   • Scans your codebase with the File Explorer agent
   • Plans changes with the Thinker agent
   • Makes edits with the Editor agent
   • Reviews results with the Reviewer agent

3. Review the changes shown in the terminal

Step 3: Use Agent Modes

Codebuff supports different modes for different needs:

Mode	Flag	Description	Best For
Default	(none)	Balanced performance	Most tasks
Free	--free	Cost-optimized	Simple tasks
Max	--max	Maximum capability	Complex refactors
Plan	--plan	Planning only	Architecture planning

# Use max mode for complex tasks
codebuff --max

# Use free mode for simple changes
codebuff --free

# Planning mode
codebuff --plan

Step 4: Use Slash Commands

Inside the CLI, use these commands:

Command	Description
/init	Initialize agent configuration
/help	Show available commands
/image	Analyze an image
/referral	View referral code
/usage	Check credit usage

Step 5: Continue Previous Sessions

# Continue last conversation
codebuff --continue

# Continue a specific conversation
codebuff --continue <conversation-id>

(cli/src/index.tsx:100-152)

Expected Results

• ✅ Terminal UI renders with chat interface
• ✅ Agent processes your request and makes file changes
• ✅ Changes are applied directly to your codebase
• ✅ Summary of all modifications displayed

Troubleshooting

🔴 Error: Authentication required

Cause: No valid auth token found.

Solution:

1. Run codebuff login to authenticate via GitHub
2. Or provide an API key: codebuff --api-key=your-key

Source: (cli/src/index.tsx:189-192)

🔴 Error: Credit limit reached

Cause: Insufficient credits for the operation.

Solution:

1. Check usage: /usage command
2. Upgrade your plan at codebuff.com/pricing
3. Enable auto top-up in your profile settings

FAQ

Q: Which LLM models does Codebuff use?

A: Codebuff uses OpenRouter to access any model — Claude, GPT, DeepSeek, Qwen, and more. The model is selected based on your agent mode (Free/Default/Max/Plan).

Q: Can I use my own API key?

A: Yes, Codebuff supports Bring-Your-Own-Key (BYOK) for Anthropic, OpenAI, and Google Gemini. Configure in your profile settings.

Related

• Custom Agents — Create your own agents
• SDK Integration — Use Codebuff programmatically
• Architecture — How the multi-agent system works