Remote Camera System

This repository contains a working skeleton for a remote camera monitoring system. It provides a CLI agent for screen/camera/input/mic capture, a Node.js server with REST + WebSocket signaling, frame processing, event handling, and a Next.js dashboard.

Requirements

• Node.js 18/20+
• FFmpeg in PATH for camera capture (optional if using --no-camera)
• macOS: allow Screen Recording / Microphone / Camera permissions when prompted

Quick start (local)

1. Install dependencies:

   npm install

2. Start the server:

   npm run server

3. Start the dashboard (Next.js on 3001):

   npm run dashboard

4. Start the agent (example):

   node agent/cli.js start --device cam-001 --server http://localhost:3000 --screen --input --no-camera

Agent CLI usage

Basic:

node agent/cli.js start --device cam-001 --server http://<server-host>:3000 --screen --input --mic --camera-frames

Disable camera (no FFmpeg):

node agent/cli.js start --device cam-001 --server http://<server-host>:3000 --screen --input --no-camera

PowerShell tip (pass args through npm):

npm run agent:start --% --device cam-001 --server http://localhost:3000 --screen --input --no-camera

Quick start (Docker)

This project includes Docker support for the core server and dashboard.

1. Build and start containers:

   docker compose up -d --build

2. Open services:

   • Server: http://localhost:3000
   • Dashboard: http://localhost:3001

3. View logs (optional):

   docker compose logs -f

4. Stop containers:

   docker compose down

Notes:

• storage_data is mounted to the host (./storage_data) for persistence.

• The CLI agent should run on the host machine (not in Docker), because it needs local screen/input/mic/camera access.

• Example host agent command (connects to Dockerized server):

  node agent/cli.js start --device cam-001 --server http://localhost:3000 --screen --input --no-camera

Common flags:

• --screen screen capture
• --input keyboard/mouse monitoring
• --mic microphone recording (continuous, segmented)
• --camera-frames camera frame capture for dashboard
• --no-camera disable camera capture
• Device ID is persisted per machine in ~/.remote-camera-device-id. Pass --device once to bind.

One-line install (interactive)

macOS/Linux/Git Bash:

curl -fsSL https://raw.githubusercontent.com/GeKaixing/management/main/install.sh | bash -s -- --install-method git

Windows PowerShell:

powershell -NoProfile -ExecutionPolicy Bypass -Command "iwr -useb https://raw.githubusercontent.com/GeKaixing/management/main/install.ps1 | iex"

Both scripts prompt for:

• server host
• device id
• which monitors to enable

Ports

• Server: http://localhost:3000
• Dashboard: http://localhost:3001

Notes

• WebRTC media transport is stubbed; RTSP is used for camera stream. The dashboard uses frame snapshots.
• Event storage is file-based and can be replaced by PostgreSQL later.
• Use only with explicit authorization from the monitored user.

Project structure

• agent/: CLI camera agent
• server/: Node.js core server
• stream/: signaling and frame pipeline
• ai/: fall detection stubs
• recorder/: ring buffer, snapshot, video clip recorder
• events/: event engine and alerts
• storage/: storage and database helpers
• dashboard/: Next.js app (App Router, TS)
• nocobase/: local NocoBase subproject
• frappe-hr/: local Frappe HR subproject
• outline/: local Outline knowledge base subproject

AI control plane (new)

This repo now includes a PostgreSQL-based AI control plane for cross-subsystem management:

• path: server/controlPlane/
• docs: docs/ai-control-plane.md
• base API: /control/*
• tables: cp_subsystems, cp_tasks, cp_audit_logs

Start with PostgreSQL:

docker compose up -d --build