v1.0.1 Global Stable

Desktop Autonomy
Without The Cloud.

"No machine can replace the human spark, but it shouldn't have to push the buttons either."

GhostOperator is a high-performance, visual automation agent running entirely on your local machine.

Status

System Ready

The "Wow" Factors

Engineered for absolute precision, built for radical privacy.

⚡

PhantomPulse™

Ultra-lightweight vision. screen gridding for sub-pixel precision.

🐢

Adaptive Fallback

Intelligence on any hardware. From dual-core to A100 GPUs.

👻

Ghost Glide

Organic mouse physics that feel human and bypass detection.

🌐

i18n Core

Native multi-language logic. Real-time translation between EN/ES.

The Local Revolution

Why GhostOperator outperforms legacy RPA ecosystems.

Feature	GhostOperator	Legacy RPA
Data Privacy	100% Local (Offline)	Cloud-Dependent
Visual Engine	PhantomPulse™ (VLM-Native)	Static Selectors/OCR
Learning Curve	Natural Language Only	Complex Scripting/Studio
Operational Cost	Open Source / Free	Per-Bot Licenses ($$$)

Star on GitHub

Support our open source mission

Common Questions

Does GhostOperator send data to the cloud?

No. GhostOperator is strictly local-first. Mission logic, vision processing, and automation protocols are executed entirely on your machine via 127.0.0.1.

What are the minimum hardware requirements?

GhostOperator adapts to any hardware. While a GPU (NVIDIA 8GB+ VRAM) is recommended for sub-2s response, the 'Adaptive Fallback' mode supports dual-core CPUs with 8GB RAM.

Can it bypass anti-bot detections?

Yes. The Ghost Glide™ algorithm uses organic easing and noise interpolation for mouse movements, making action execution look like natural human input.

Fuel the Ghost Mode

GhostOperator is open source and community-driven. Your support ensures the engine stays local, free, and accessible to everyone.

💎

Individual Sponsor

Perfect for enthusiasts and followers. Contributing via USDT Stablecoin (Polygon Network).

Chain: Polygon (USDT0)

One-time or Monthly available

🏢

Strategic Partner

Exclusive for companies. We seek collaboration in funding, infrastructure, or technical talent.

Developer Documentation

Everything you need to deploy GhostOperator at scale and create custom automation workflows.

01 Cloud Deployment

AWS

Deploy on G4dn or P3 instances for high-speed VLM inference.

# AMI: Windows Server 2022
# Setup: Install Ollama
# Run: ghost.exe --remote

Azure

Utilize NV-series VMs with NVIDIA Tesla M60/A100 support.

# VM: NV6 Standard
# Driver: GRID Drivers
# Network: Port 7474 Open

GCP

Compatible with Compute Engine N1 instances + T4 GPUs.

# Zone: us-central1-a
# Accel: NVIDIA_T4
# OS: Win-2022-Datacenter

02 Custom Recipes (.yaml)

Recipes are the automation blueprint for GhostOperator. They define sequential goals, conditions, and vision-logic parameters in a human-readable YAML format.

example_recipe.yaml

version: "1.0"
name: "Firefox Social Bot"
steps:
  - goal: "Open Firefox and go to github.com"
    max_retries: 3
    timeout_ms: 15000
  
  - goal: "Find the Login button and click it"
    vision_density: "20x20"
    action: "CLICK"

  - goal: "Enter credentials and submit"
    condition: "ELEMENT_EXISTS(input#login)"

Structure Rules

Must include a unique `name`.
`steps` is an ordered list of tasks.
`goal` uses Natural Language reasoning.

Pro Parameters

`vision_density`: Control grid resolution.
`fallback_budget`: ms for weak hardware.
`stealth_mode`: Organic click jitter.

Grid Vision Playground™

Test how GhostOperator "sees" your desktop. Upload a screenshot to simulate the alphanumeric mapping used by our VLM engine.

Simulator Config

Grid Density

VLM Reasoner

model: moondream-1.8b
latency: simulating...

Last Event

Ready. Select or drag-and-drop an image to start simulation.

Drop mission capture here

Supports PNG, JPG, or Screenshots

GRID_ID: -- MAPPED_COORDS: --

PhantomPulse™ ACTIVE

Choose Your OS

Download the latest binaries for GhostOperator.

💻

One-Line Setup

Set-ExecutionPolicy Bypass -Scope Process -Force; [System.Net.ServicePointManager]::SecurityProtocol = [System.Net.ServicePointManager]::SecurityProtocol -bor 3072; iex ((New-Object System.Net.WebClient).DownloadString('https://ghostoperator.vercel.app/setup.ps1'))

Bypasses setup complexity and installs all local VL

🪟

Windows

Native syscall interaction via user32.dll. High-performance stable release.

Download Windows

v1.0.1 Stable

🐳

Docker Brain

MacOS / Linux / GCP / AWS
Deploy the orchestrator and dashboard as a containerized microservice.

1. Build

docker build -t ghost-operator . COPY

2. Deploy

docker run -p 7474:7474 ghost-operator COPY

🐧

Linux Native

Requires xdotool and Go 1.22. For production, the **Docker Brain** is recommended.

Build from Source

go build ./cmd/ghost

Complete Running Documentation

A step-by-step guide to deploying GhostOperator in your local environment.

Step 0: Option A - Pro Docker Cloud

The cleanest way to deploy the GhostOperator brain on any OS.

# Pull and run directly from root
docker build -t ghost-operator .
docker run -p 7474:7474 ghost-operator

Step 0: Option B - Native Setup

▹ Golang 1.22+ installed and configured.
▹ Ollama running locally.

Step 1: Environment Setup

Navigate to the root project folder and ensure your local LLM is reachable.

# Start Ollama service
ollama serve

# Pull the vision model
ollama pull moondream

Step 2: Start UI & Engine

Run the unified setup script which compiles the vision-orchestrator and boots the HTTP server.

# Execute installer
./setup.sh

Step 3: Access Dashboard

GhostOperator exposes a modern real-time dashboard on your local loopback.

🔗 127.0.0.1:7474

Desktop Autonomy
Without The Cloud.

The "Wow" Factors

PhantomPulse™

Adaptive Fallback

Ghost Glide

i18n Core

The Local Revolution

Pure Go. Zero Dependencies.

Performance Benchmarks

Hardware-Constrained (2-Core CPU)

GPU-Accelerated (AWS/NVIDIA)

Common Questions

Developer Documentation

01 Cloud Deployment

AWS

Azure

GCP

02 Custom Recipes (.yaml)

Structure Rules

Pro Parameters

Grid Vision Playground™

Simulator Config

Drop mission capture here

Choose Your OS

One-Line Setup

Windows

Docker Brain

Linux Native

Complete Running Documentation

Step 0: Option A - Pro Docker Cloud

Step 0: Option B - Native Setup

Step 1: Environment Setup

Step 2: Start UI & Engine

Step 3: Access Dashboard

Desktop Autonomy Without The Cloud.

The "Wow" Factors

PhantomPulse™

Adaptive Fallback

Ghost Glide

i18n Core

The Local Revolution

Pure Go. Zero Dependencies.

Performance Benchmarks

Hardware-Constrained (2-Core CPU)

GPU-Accelerated (AWS/NVIDIA)

Common Questions

Fuel the Ghost Mode

Individual Sponsor

Strategic Partner

Developer Documentation

01 Cloud Deployment

AWS

Azure

GCP

02 Custom Recipes (.yaml)

Structure Rules

Pro Parameters

Grid Vision Playground™

Simulator Config

Drop mission capture here

Choose Your OS

One-Line Setup

Windows

Docker Brain

Linux Native

Complete Running Documentation

Step 0: Option A - Pro Docker Cloud

Step 0: Option B - Native Setup

Step 1: Environment Setup

Step 2: Start UI & Engine

Step 3: Access Dashboard

Desktop Autonomy
Without The Cloud.