Agent Coder 🤖

A production-ready AI coding agent with interactive terminal interface, built with Vercel AI SDK and TypeScript.

Features

• 🎨 Modern terminal interface with chalk and ora
• 💬 Interactive conversation with the agent
• 🔧 Tool execution with confirmation for dangerous operations
• 📝 Type-safe tool definitions with Zod
• 🎯 Streaming responses from LLM providers
• 🔒 Sandboxed workspace for safe file operations
• ⚙️ Configurable via environment variables
• 🧪 TypeScript with full type safety

Requirements

• Node.js 20+
• OpenAI API Key or Anthropic API Key

Quick Start

1. Clone the repository

git clone git@github.com:kikeztw/agent-coder.git
cd agent-coder

2. Install dependencies

npm install

3. Configure environment variables

Create a .env file in the project root:

OPENAI_API_KEY=your-api-key-here
# Or if using Anthropic:
# ANTHROPIC_API_KEY=your-api-key-here

4. Build the project

npm run build

5. Link the CLI globally

npm link

Usage

Once linked, you can run the agent from any directory:

coder chat

Available Commands

• coder chat - Start interactive session
• coder run "your message" - Run single query
• exit or quit - Exit interactive session

Usage Examples

👤 You: Create a TypeScript file that calculates Fibonacci numbers

👤 You: Explain what the main.ts file does

👤 You: Refactor the workflow function to improve readability

Development

Development Setup

# Install dependencies
npm install

# Run in watch mode
npm run dev

# Type checking
npm run typecheck

# Run tests
npm test

Project Structure

agent-coder/
├── src/
│   ├── core/           # Core agent logic
│   │   ├── state.ts    # Agent state types
│   │   └── workflow.ts # ReAct workflow
│   ├── nodes/          # Workflow nodes
│   │   ├── thought.ts  # LLM reasoning
│   │   ├── action.ts   # Tool execution
│   │   └── review.ts   # Human confirmation
│   ├── tools/          # Tool implementations
│   │   ├── registry.ts # Tool registration
│   │   ├── file-tools.ts
│   │   └── execution-tools.ts
│   ├── providers/      # LLM providers
│   │   └── llm.ts
│   ├── config/         # Configuration
│   │   └── settings.ts
│   ├── cli/            # CLI interface
│   │   └── index.ts
│   └── utils/          # Utilities
│       └── security.ts
├── tests/              # Test suite
├── workspace/          # Agent sandbox directory
├── package.json
├── tsconfig.json
└── README.md

Available Tools

The agent has access to the following tools:

• read_file: Read file contents
• write_file: Create or overwrite files
• delete_file: Delete files (requires confirmation)
• list_directory: List directory contents
• run_command: Execute shell commands (requires confirmation)

All tools operate within the sandboxed workspace/ directory for security.

Configuration

Configuration is managed through environment variables:

• Model Configuration: MODEL_PROVIDER, MODEL_NAME, TEMPERATURE
• API Keys: OPENAI_API_KEY or ANTHROPIC_API_KEY
• Security: DANGEROUS_TOOLS, MAX_EXECUTION_TIME
• Workspace: WORKSPACE_PATH

License

MIT License - see LICENSE file

Author

kikeztw - GitHub