AtomCode Daemon

HTTP + SSE API service for AtomCode AI agent with file operation tools.

Prerequisites

• Node.js >= 18.0.0
• AtomCode CLI — installed and logged in (atomcode login)
• config.toml — auto-generated at ~/.atomcode/config.toml after first login

Quick Start

# 1. Clone the repository
git clone https://github.com/Rosenbad/atomcode-daemon.git
cd atomcode-daemon

# 2. Install dependencies
npm install

# 3. Start the server
npm start

You'll see:

  🌐  http://127.0.0.1:18788
  🔧  CLI available: true

Open http://127.0.0.1:18788 in your browser to use the web UI.

How It Works

1. The daemon reads your AtomCode configuration from ~/.atomcode/config.toml automatically.
2. It spawns the atomcode CLI in headless mode (atomcode -p "...") for AI responses.
3. All providers defined in your config.toml appear in the UI dropdown.
4. The CLI has built-in file tools — no separate tool server is needed.
5. Sessions and messages are stored locally in data/store.json.

Configuration

Create or edit ~/.atomcode/config.toml:

default_provider = "openai"

[providers.openai]
type = "openai"
api_key = "sk-your-key"
model = "gpt-4o"
base_url = "https://api.openai.com/v1"

[daemon]
host = "127.0.0.1"
port = 18788

The daemon hot-reloads config.toml when it changes — no restart needed.

Features

• HTTP API with SSE streaming chat
• Web UI for browser-based interaction
• CLI Mode — uses atomcode CLI headless mode with built-in file tools
• API Mode — direct OpenAI-compatible API calls for Vision models
• Image Analysis — vision API support
• Session Management — persistent conversation history (local)
• Authentication — optional token-based auth
• Directory Switching — change working directory from UI or /cd endpoint

API Endpoints

Chat

• POST /chat — SSE streaming chat
• POST /chat/stop — Stop ongoing chat

Sessions

• GET /sessions — List sessions
• POST /sessions — Create new session
• GET /sessions/search?q=query — Search sessions

Tools

• GET /tools — List available tools
• GET /tools/file/read?path=... — Read file
• POST /tools/file/write — Write file
• POST /tools/file/edit — Search and replace
• GET /tools/dir/list — List directory
• POST /tools/bash — Execute command
• GET /tools/grep?pattern=... — Search in files
• GET /tools/glob?pattern=... — Find files

System

• GET /health — Health check
• GET /models — List available models
• POST /config/reload — Reload configuration
• POST /cd — Change working directory

Authentication

Enable authentication by setting environment variables:

ATOMCODE_AUTH_ENABLED=true
ATOMCODE_ADMIN_KEY=your-admin-key

Then use the token in requests:

curl -H "Authorization: Bearer <token>" http://localhost:18788/tools/file/read?path=README.md

Important Notes

• The daemon must be running while you use the web UI.
• The data/ directory stores session history. Deleting it resets all conversations.
• The daemon binds to 127.0.0.1 by default — not accessible from other machines.
• If CLI available: false, make sure atomcode is in your PATH or installed at D:\AtomCode\.
• Windows users: the daemon searches common AtomCode install paths automatically.

Project Structure

atomcode-daemon/
├── src/
│   ├── index.js          # Main server
│   ├── config.js         # Configuration management + hot reload
│   ├── auth.js           # Token authentication
│   ├── store.js          # Session storage (local JSON)
│   ├── engine.js         # AI engine (CLI + API modes)
│   ├── tools.js          # 10 file/shell operations
│   ├── toml-parser.js    # TOML config parser
│   ├── logger.js         # Structured logging
│   └── web-ui.js         # Static file serving
├── public/
│   └── index.html        # Web UI
├── data/
│   └── store.json        # Session data (generated at runtime)
├── .gitignore
├── package.json
└── README.md

Environment Variables

Variable	Description	Default
ATOMCODE_AUTH_ENABLED	Enable token authentication	false
ATOMCODE_ADMIN_KEY	Admin key for token management	-
ATOMCODE_CORS_ORIGIN	CORS origin	*
LOG_LEVEL	Log level (DEBUG, INFO, WARN, ERROR)	INFO

License

MIT