Workhorse IDE 2.0

A production-ready browser-based code editor with local GPU-accelerated AI, zero cloud dependency, complete privacy.

GitHub repository: https://github.com/paulcombs56-ops/WorkhorseIDE

Repository governance:

Branch protection enabled on main
CI required via GitHub Actions workflow CI
Topics and repository description configured

🎯 Features

Code Editor

📝 Monaco Editor - Industry-standard, VS Code quality code editing
🎨 Syntax Highlighting - JavaScript & Python (extensible)
⌨️ Smart Shortcuts - Ctrl+Enter to run, arrow keys for history
🎭 Console Themes - Dark, Neon, Hacker Green

Code Execution

⚡ JavaScript - Direct eval in browser
🐍 Python - Server-side execution with streaming output
📊 Real-time Console - Timestamped output with color coding

AI Features (Local GPU-Accelerated)

🔍 Code Analysis - Find bugs, code smells, issues
♻️ Code Refactoring - Simplify, extract functions, optimize
📚 Documentation - Auto-generate docstrings (Google/NumPy/Sphinx)
🎯 Code Completion - LLM-powered suggestions (extensible)
✨ Code Formatting - Prettier (JS) + Black (Python)

Architecture

Frontend: React-ready, Vanilla JS (upgradable)
Backend: Express.js + Python FastAPI
AI Engine: Ollama or vLLM
GPU Support: NVIDIA CUDA, AMD ROCm, Intel Ryzen AI NPU

🚀 Quick Start (30 minutes)

The former phone-and-desktop assistant has been extracted into the standalone pulse-assistant project.

Prerequisites

Node.js 18+
Python 3.8+
8GB+ VRAM (16GB recommended)
GPU (NVIDIA RTX 3060+, AMD Radeon RX 5700 XT+, or Ryzen AI)

Setup

# 1. Install Ollama
# Download from https://ollama.ai

# 2. Pull a model
ollama pull codellama:7b

# 3. In Terminal 1: Start Ollama
ollama serve

# 4. In Terminal 2: Start Node backend
cd backend
npm install
node app.js

# 5. In Terminal 3: Start Python backend
cd ai_backend
pip install -r requirements.txt
python main.py

# 6. In Terminal 4: Open IDE
# http://localhost:3001/index.html

Verify Setup

# Run verification script
./verify-services.sh          # Linux/Mac
./verify-services.bat         # Windows

Or manually test:

curl http://localhost:3001/api/hello      # Node backend
curl http://localhost:8888/health         # Python backend
curl http://localhost:11434/api/tags      # Ollama

📚 Documentation

Document	Purpose
QUICKSTART.md	Step-by-step setup + 6 workflow examples
ARCHITECTURE.md	System architecture + checklist + verification
AI_INTEGRATION.md	What's new + how to use AI features
ai_backend/README.md	API reference + troubleshooting
ai_backend/SETUP.md	GPU optimization for NVIDIA/AMD/Ryzen AI

🎮 Using the IDE

Buttons

Button	Shortcut	Purpose
⚡ Run	Ctrl+Enter	Execute JS/Python
✨ Format	-	Auto-format code
🔍 Analyze	-	Find issues (AI)
♻️ Refactor	-	Improve code (AI)
📚 Generate Docs	-	Add docstrings (AI)
🤖 AI Status	-	Check AI backend

Example: Analyze Python Code

# Write in editor
def buggy_code():
    x = 10
    print(undefined_var)  # Oops!
    return result        # Result not defined

# Click "🔍 Analyze"
# Output: Shows undefined variables and issues

Example: Refactor JavaScript

// Write in editor
if (x === true) {
    console.log("ok");
}

// Click "♻️ Refactor" → choose "simplify"
// Output: 
if (x) {
    console.log("ok");
}

📁 Project Structure

Workhorse 2.0/
├── frontend/                  # Browser IDE
│   ├── index.html
│   ├── main.js               # Thin bootstrap/orchestration
│   ├── aiService.js          # AI API client + chat transport
│   ├── chatModule.js
│   ├── chatRenderModule.js
│   ├── aiFeaturesModule.js
│   ├── consoleModule.js
│   ├── editorModule.js
│   ├── executionModule.js
│   ├── explorerModule.js
│   ├── panelResizerModule.js
│   └── styles.css
├── backend/                   # Node.js server
│   ├── app.js                # Express route composition
│   ├── routes/               # health/explorer/exec/ai/providerAi
│   ├── services/             # formatter/executor/providers/chain
│   ├── middleware/
│   ├── utils/
│   └── config.js
│   └── package.json
├── ai_backend/               # Python AI server
│   ├── main.py
│   ├── requirements.txt
│   ├── .env.example
│   ├── README.md
│   └── SETUP.md
├── QUICKSTART.md             # ← Read this first!
├── ARCHITECTURE.md
├── AI_INTEGRATION.md
├── verify-services.sh        # Test everything works
└── verify-services.bat       # Windows version

🛠️ API Endpoints

Node backend runs on http://localhost:3001.

Legacy AI proxy (deprecated, still available by default):

GET /ai/health
POST /ai/analyze
POST /ai/refactor
POST /ai/generate-docs
POST /ai/complete
GET /ai/models

Provider chain route (new):

GET /api/ai/health
POST /api/ai
Body: {prompt, model?}
→ {response, source, model, ...}

Legacy route details:

# Health check
GET /health
→ {status, provider, model}

# Analyze code
POST /analyze
Body: {code, language}
→ {issues[], suggestions[]}

# Refactor code  
POST /refactor
Body: {code, language, refactor_type}
→ {refactored_code, explanation}

# Generate documentation
POST /generate-docs
Body: {code, language, style}
→ {documentation}

# Code completion
POST /complete
Body: {code, language, cursor_position, context_lines}
→ {completion, confidence}

# List models
GET /models
→ {provider, models[]}

⚙️ Configuration

Edit ai_backend/.env:

# AI provider: "ollama" or "vllm"
AI_PROVIDER=ollama

# Ollama URL
OLLAMA_URL=http://localhost:11434

# Default model
DEFAULT_MODEL=codellama:7b

# Backend port
AI_BACKEND_PORT=8888

Backend security (optional) is configured via environment variables when starting backend/app.js:

# Backend HTTP port
PORT=3001

# Comma-separated CORS allowlist
ALLOWED_ORIGINS=http://localhost:3001,http://127.0.0.1:3001

# Optional API token to protect sensitive routes
WORKHORSE_API_TOKEN=replace-with-a-strong-token

# Legacy /ai/* compatibility proxy (deprecated)
ENABLE_LEGACY_AI_PROXY=true
LEGACY_AI_PROXY_SUNSET=2026-07-31

# Provider-chain default model
WORKHORSE_DEFAULT_CHAT_MODEL=llama3

When WORKHORSE_API_TOKEN is set, the frontend should provide the same token in browser local storage:

localStorage.setItem("workhorse-api-token", "replace-with-a-strong-token");

Use Your Own Anthropic Key (Claude)

Open the IDE and set chat Route to Provider.
Set Provider to Claude (or keep Auto to allow fallback).
Paste your key into Anthropic API key and click Save Key.

The key is stored only in your browser local storage and is sent as x-anthropic-api-key to the local Node backend.

🎓 Supported Models

Model	VRAM	Speed	Quality	Command
CodeLlama 7B	~5GB	Fast	Excellent	`ollama pull codellama:7b`
CodeLlama 13B	~12GB	Medium	Better	`ollama pull codellama:13b`
Mistral 7B	~5GB	Fast	Good	`ollama pull mistral:7b`
Neural Chat 7B	~5GB	Fast	Good	`ollama pull neural-chat:7b`
CodeLlama 7B (Q4)	~3GB	Fast	Decent	`ollama pull codellama:7b-q4`

Recommended: CodeLlama 7B for most users

🚦 Performance

GPU	Model	Latency
RTX 4090	CodeLlama 7B	<1s
RTX 4070	CodeLlama 7B	2-4s
RTX 4060	CodeLlama 7B	4-8s
RTX 3060 12GB	Mistral 7B	5-10s
Recommended Setup	CodeLlama 7B	4-8s

🔒 Privacy & Security

✅ Local-First - Core IDE and local model flow stay on your machine ✅ Optional Cloud Providers - Claude/Groq/OpenAI can be enabled by your keys ✅ No Tracking - No telemetry or logging ✅ Open Source - Inspect & modify the code ✅ Provider Choice - Use local-only or hybrid fallback as needed

🐛 Troubleshooting

AI Features Show "Offline"

# Ensure all 3 services are running:
ollama serve                    # Terminal 1
cd backend && node app.js       # Terminal 2  
cd ai_backend && python main.py # Terminal 3

Slow AI Responses (>10 seconds)

# Check GPU usage
nvidia-smi -l 1

# Reduce model size
ollama pull mistral:7b
# or quantized
ollama pull codellama:7b-q4

Out of Memory

# Use quantized version (-q4, -q5)
ollama pull codellama:7b-q4

# Or smaller model
ollama pull neural-chat:7b

Port Already in Use

# Find & kill process on port
lsof -i :3001                           # Mac/Linux
netstat -ano | findstr :3001            # Windows
kill <PID>                              # Mac/Linux
taskkill /PID <PID> /F                  # Windows

See QUICKSTART.md for detailed troubleshooting.

🔄 Workflow Examples

Example 1: Clean & Document Code

Write code → ✨ Format → 🔍 Analyze → 📚 Generate Docs

Example 2: Optimize Function

Write slow function → 🔍 Analyze → ♻️ Refactor (optimize) → ⚡ Run to verify

Example 3: Debug Code

Run code (⚡) → See error → 🔍 Analyze → Fix → ⚡ Run again

📈 Tech Stack

Layer	Technology	Purpose
Frontend	Monaco Editor, Vanilla JS	Code editing & UI
Backend	Express.js	HTTP server + proxy
AI	FastAPI (Python)	AI service endpoints
Inference	Ollama / vLLM	Local LLM serving
GPU	CUDA / ROCm / Ryzen AI	Hardware acceleration

🎯 Next Steps

✅ Run Setup - Follow QUICKSTART.md
🧪 Test Services - Run verify-services.sh (or .bat on Windows)
💡 Try Features - Click all AI buttons
⚡ Optimize - Read ai_backend/SETUP.md for GPU tuning
🚀 Deploy - Run as background service (systemd, Docker, etc.)

🤝 Contributing

Contributions welcome! Areas for enhancement:

Real-time code completion as-you-type
Streaming responses from AI backend
Model switcher UI
Project-wide code analysis
Keyboard shortcuts for AI features
Dark mode improvements

📝 License

MIT - Use freely in personal & commercial projects

🆘 Support

📖 Read QUICKSTART.md for setup help
🏗️ Check ARCHITECTURE.md for system overview
✅ Use docs/FRONTEND_REGRESSION_CHECKLIST.md after frontend changes
🔧 See ai_backend/SETUP.md for GPU troubleshooting
🤖 See ai_backend/README.md for API reference

💡 FAQ

Q: Does this send my code to the cloud? A: No. Everything runs locally. No data leaves your machine.

Q: What GPUs are supported? A: NVIDIA (CUDA 12.x), AMD (ROCm), Intel Ryzen AI NPU. Check ai_backend/SETUP.md.

Q: Can I use this without a GPU? A: Not recommended - CPU inference is 100x slower. GPUs are essential.

Q: What models can I use? A: Any model available in Ollama. CodeLlama 7B recommended for code tasks.

Q: Why is it slow the first time? A: Model is loading from disk to VRAM. Subsequent calls are much faster.

Q: Can I add new AI features? A: Yes. Add backend endpoints and wire UI handlers in the relevant frontend modules (for example aiFeaturesModule.js and chatModule.js), keeping main.js as bootstrap-only orchestration.

Q: How do I switch models? A: ollama pull mistral:7b, then set DEFAULT_MODEL=mistral:7b in .env.

🎉 Ready to Code?

# 1. Follow QUICKSTART.md
# 2. Run verify-services.sh/bat
# 3. Open http://localhost:3001/index.html
# 4. Click 🤖 AI Status to verify
# 5. Start coding!

Enjoy your local, private, GPU-accelerated IDE! 🚀

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
.continue		.continue
.github		.github
ai_backend		ai_backend
backend		backend
calculator		calculator
docs		docs
electron		electron
frontend		frontend
pulse-assistant		pulse-assistant
sandbox		sandbox
.aider.conf.yml		.aider.conf.yml
.gitignore		.gitignore
AI_INTEGRATION.md		AI_INTEGRATION.md
ARCHITECTURE.md		ARCHITECTURE.md
DEVELOPER.md		DEVELOPER.md
IMPLEMENTATION_SUMMARY.md		IMPLEMENTATION_SUMMARY.md
LOCAL_AGENT_SETUP.md		LOCAL_AGENT_SETUP.md
QUICKSTART.md		QUICKSTART.md
README.md		README.md
SETUP_CHECKLIST.md		SETUP_CHECKLIST.md
calculator.html		calculator.html
email.html		email.html
forge.config.js		forge.config.js
package-lock.json		package-lock.json
package.json		package.json
start-local-agent.ps1		start-local-agent.ps1
start-workhorse.ps1		start-workhorse.ps1
verify-agent-endpoints.ps1		verify-agent-endpoints.ps1
verify-ai-smoke.ps1		verify-ai-smoke.ps1
verify-services.bat		verify-services.bat
verify-services.sh		verify-services.sh

Folders and files

Latest commit

History

Repository files navigation

Workhorse IDE 2.0

🎯 Features

Code Editor

Code Execution

AI Features (Local GPU-Accelerated)

Architecture

🚀 Quick Start (30 minutes)

Prerequisites

Setup

Verify Setup

📚 Documentation

🎮 Using the IDE

Buttons

Example: Analyze Python Code

Example: Refactor JavaScript

📁 Project Structure

🛠️ API Endpoints

⚙️ Configuration

Use Your Own Anthropic Key (Claude)

🎓 Supported Models

🚦 Performance

🔒 Privacy & Security

🐛 Troubleshooting

AI Features Show "Offline"

Slow AI Responses (>10 seconds)

Out of Memory

Port Already in Use

🔄 Workflow Examples

Example 1: Clean & Document Code

Example 2: Optimize Function

Example 3: Debug Code

📈 Tech Stack

🎯 Next Steps

🤝 Contributing

📝 License

🆘 Support

💡 FAQ

🎉 Ready to Code?

About

Topics

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages