RDC - Remote Dev Ctrl

Monitor and control your AI coding agents from anywhere - phone, tablet, or browser.

RDC is an open-source command center for developers running multiple AI agents (Claude Code, Cursor, Gemini CLI) across multiple projects. One dashboard, all your terminals, accessible from any device.

Why RDC?

• See all your agents in one place - 5 projects with Claude Code running? See them all, switch instantly
• Check on agents from your phone - Mobile-first UI with terminal access, virtual keyboard, and touch controls
• Shared browser preview - You and the AI agent see the same page. Click, type, navigate together
• Terminals that survive everything - PTY sessions persist across server restarts and network drops

Quick Start

# Install with uv (recommended — isolated, auto-managed CLI)
uv tool install rdc

# Or with pip
pip install rdc

rdc setup          # Guided config: API keys, presets, remote access
rdc server start   # Open http://localhost:8420

One-off run without installing:

uvx rdc setup

Or install from source:

# One-line install (clones, installs deps, builds, runs guided setup)
curl -sSL https://raw.githubusercontent.com/dawalama/remote-dev-control/main/install.sh | bash

# Or manual:
git clone https://github.com/dawalama/remote-dev-control.git
cd remote-dev-control
uv sync && cd frontend && pnpm install && pnpm run build && cd ..
rdc setup
rdc server start

Prerequisites

Python 3.11+, Node.js 18+, and Chrome (for browser preview).

# If you don't have uv/pnpm:
curl -LsSf https://astral.sh/uv/install.sh | sh
npm install -g pnpm

Features

Three Layouts, All First-Class

Desktop	Mobile	Kiosk
IDE-like split view	Card-based, touch-optimized	Tablet sidebar + terminal
Terminal + sidebar tabs	Fullscreen terminal overlay	Collapsible side panels

Terminal Management

• Spawn Claude Code, Cursor, Gemini CLI, or plain shell per project
• WebSocket streaming with auto-reconnect and snapshot-based session restore
• Terminal switcher: tap title or long-press Back to switch between terminals
• Virtual keyboard with arrow keys, Ctrl-C, Tab, Enter for mobile
• Waiting-for-input detection with visual alerts

Browser Automation

• Local Chrome (no Docker required) with CDP screencast viewer
• 5 agent tools: observe, click, type, navigate, screenshot
• Observe-act loop: agent sees the page, decides actions, executes, verifies
• Same page shared between you and the agent - no separate sessions

Actions (Services & Commands)

• Auto-discover project scripts from package.json, Makefile, etc.
• Services (dev servers) with port detection and browser preview
• Commands (builds, tests, lints) with output capture
• Start, stop, restart, attach to orphaned processes

AI Orchestrator

Natural language control via chat:

• "Start a terminal for my-project"
• "Show running tasks"
• "Switch to kiosk mode"

Remote Access

Access from anywhere via Cloudflare Tunnel + Caddy reverse proxy:

Internet -> Cloudflare Tunnel -> Caddy (:8888) -> RDC (:8420)
                                               -> Dev servers (:3000, :5173, etc.)

Each dev server gets its own preview subdomain automatically.

Architecture

src/remote_dev_ctrl/server/
  app.py            # FastAPI app, routes, lifespan
  chrome.py         # Local Chrome process lifecycle
  browser.py        # Browser sessions, CDP connections
  browser_use.py    # Agent browser control (observe/act/screenshot)
  terminal.py       # PTY management, WebSocket relay, snapshots
  worker.py         # Task execution engine
  config.py         # Config loading
  agents/tools.py   # Agent tool definitions (file, git, browser)

frontend/src/
  layouts/          # desktop.tsx, mobile.tsx, kiosk.tsx
  features/         # terminal, browser, chat, tasks, processes
  stores/           # Zustand state management
  hooks/            # use-browser-agent, use-mount-effect

Stack: Python (FastAPI) + React + TypeScript + Tailwind + Zustand + xterm.js + SQLite

Configuration

Config lives at ~/.rdc/config.yml (auto-created on first run):

server:
  host: 127.0.0.1
  port: 8420

browser:
  backend: chrome     # Local Chrome, no Docker needed
  headless: true

providers:
  cursor:
    type: cursor-agent
    default: true
  ollama:
    type: ollama
    model: qwen3.5

API keys go in the vault, not config files:

rdc config set-secret ANTHROPIC_API_KEY sk-ant-...

CLI

rdc server start [-d]              # Start server (optionally as daemon)
rdc server stop                    # Stop server
rdc add <path> [-n name]           # Register a project
rdc list                           # List projects
rdc config set-secret KEY VALUE    # Store API key securely

API

Interactive docs at http://localhost:8420/docs when the server is running.

Key endpoints:

Endpoint	Description
WS /ws/state	Real-time state updates
WS /terminals/{id}/ws	Terminal I/O stream
POST /terminals	Spawn terminal session
POST /tasks	Create task
POST /browser/start	Start browser session
POST /browser/sessions/{id}/agent/loop	Multi-step browser agent
POST /orchestrator	Natural language command
GET /processes	List actions

MCP Integration

For AI assistants that support Model Context Protocol:

{
  "mcpServers": {
    "remote-dev-ctrl": {
      "command": "python",
      "args": ["-m", "remote_dev_ctrl.mcp.server"],
      "env": { "PYTHONPATH": "/path/to/remote-dev-ctrl/src" }
    }
  }
}

Remote Access Setup

For accessing RDC from your phone or another machine:

1. Cloudflare Tunnel - Exposes your local machine to the internet (free tier)
2. Caddy - Routes subdomains locally (auto-downloaded by RDC)

# Install and configure cloudflared
brew install cloudflared
cloudflared tunnel login
cloudflared tunnel create rdc

# Add to ~/.rdc/config.yml
caddy:
  enabled: true
  base_domain: yourdomain.com
  rdc_domain: rdc.yourdomain.com
  listen_port: 8888

# Start
cloudflared tunnel run rdc
rdc server start

Dashboard at https://rdc.yourdomain.com. Enable auth with:

rdc config set-secret RDC_SECRET_KEY $(openssl rand -hex 32)

Contributing

Contributions welcome. Please open an issue first to discuss what you'd like to change.

License

MIT