Prod_Layer

A productive layer deployed on any OS. Integrated with all AI, core memory to store context, and the ability to switch model mid-context. Seamless access to AI and information—do tasks in one go, save your toggle tax.

Future of Work & Productivity

Prod_Layer is a single layer that runs everywhere you work: browser, IDE, email. It plugs into all major AI providers with one config, keeps core memory so context is never lost, and lets you switch models mid-context (fast / powerful / reasoning) by task. Get seamless access to AI and information, do tasks in one go, and save your toggle tax—no more jumping between tabs. Overlays stay invisible in screen share (Meet, Zoom, Teams).

Features

• Smart AI Responses - Auto-detects code problems, definitions, questions, and instructions
• Human-like Typing - Natural typing speed with realistic variations
• Ultra Human Mode - Chain-of-thought code injection for coding interviews
• Screen Share Invisible - Windows excluded from screen capture (Zoom, Meet, Teams)
• Voice Input - Hold-to-talk speech recognition
• Vision Analysis - Screenshot + AI analysis
• Persistent Chat - Continue conversations across sessions with full history
• Cross-Model Memory - Context shared across different AI models and windows
• Multi-LLM Provider System - Seamless switching between 12+ AI providers
• Smart Routing - Automatic model selection based on task complexity
• Per-Window Context - Separate memory for each application window

Quick Start

# Install dependencies
npm install
pip install -r requirements.txt

# Configure API keys
cp config.example.env .env
# Edit .env with your keys

# Run
npm start

Keyboard Shortcuts

Shortcut	Action
Ctrl+Shift+D	Process clipboard (code/question/text)
Ctrl+Shift+P	Inject AI response
Ctrl+Alt+Enter	Fix grammar / continue writing
Ctrl+Shift+F	Screenshot + Vision analysis
Ctrl+Shift+E	Save screenshot of current window (use when OS capture is blocked)
Ctrl+Shift+V	Voice input (hold to talk)
Ctrl+Shift+S	Open settings
Ctrl+Shift+C	Open chat window
Ctrl+Shift+Space	Toggle overlay window
Escape	Cancel / close popup

Settings

Access with Ctrl+Shift+S:

Setting	Description
AI Assistant	Master on/off toggle
Dark Mode	UI theme
Human-like Typing	Natural speed variation
Auto Inject	Skip suggestion popup
Live Mode	Auto-suggest on typing pause
Coding Mode	Show code + explanation windows
Ultra Human Typing	Chain-of-thought code injection
LLM Provider	Select default AI model
API Key Management	Configure and test provider keys

Modes

Smart Clipboard (Ctrl+Shift+D)

Auto-detects content type and responds appropriately:

You Copy	AI Response
"two sum problem"	Clean code solution
"OOP"	25-35 word definition
"What is REST?"	Direct 2-4 sentence answer
Code + "explain this"	Follows your instruction

Ultra Human Typing (Interview Mode)

For coding interviews - types code like a real developer:

1. Enable Ultra Human Typing in Settings
2. Copy the coding problem
3. Click in your code editor
4. Press Ctrl+Shift+P
5. AI generates and types code with:
   • Main skeleton first with pass
   • Navigates up to add helper functions
   • Returns to implement main
   • Realistic typos and corrections
   • Variable pauses between sections

Other Modes

• Grammar Mode - Fix spelling/grammar (Ctrl+Alt+Enter on typed text)
• Extension Mode - Continue writing (Ctrl+Alt+Enter again within 2s)
• Vision Mode - Analyze screenshots (Ctrl+Shift+F)
• Voice Mode - Speech to text (Ctrl+Shift+V hold to talk)
• Chat Mode - Persistent conversations with full history (Ctrl+Shift+C)

Advanced Features

Multi-LLM Provider System

Prod_Layer supports 12+ AI providers through LiteLLM:

• Fast Models: Groq Llama 3.3 70B, Mistral Small, GPT-4o Mini, Gemini 2.0 Flash
• Powerful Models: DeepSeek Chat, Claude Sonnet, GPT-4o, Grok 4, Together Llama 3.3 70B
• Reasoning Models: DeepSeek Reasoner, Perplexity Sonar Pro, Cohere Command R+, Replicate Llama 405B

Smart Routing

• Auto Mode: Automatically selects the best model based on task complexity
• Manual Selection: Choose specific models from the settings dropdown
• Fallback System: Seamlessly switches providers on rate limits or auth errors

Memory System

• Per-Window Memory: Each application maintains separate conversation context
• Cross-Model Memory: Chat and prompt bar interactions are shared across models
• Persistent Storage: Conversations saved to disk and restored across sessions
• Context Window Management: Automatically caps history sent to LLMs to avoid overflow

Chat Window

Access with Ctrl+Shift+C:

• Full conversation history with timestamps
• Model selection dropdown
• Markdown formatting support
• Streaming responses
• Search and filter past conversations

Project Structure

NXlayer/
├── main.js                              # Electron main process
├── package.json                         # Node dependencies
├── requirements.txt                     # Python dependencies
├── .env                                 # API keys (not committed)
├── config.example.env                   # Example config
│
├── src/
│   ├── electron/
│   │   └── keystore.js                  # Encrypted API key storage
│   │
│   ├── services/                        # Python backend services
│   │   ├── ai/
│   │   │   ├── backend.py               # Persistent AI backend with streaming
│   │   │   └── prompts.py               # Context-aware smart prompting
│   │   ├── providers/                   # Multi-LLM provider system
│   │   │   ├── __init__.py              # ProviderManager interface
│   │   │   ├── router.py                # Model routing and discovery
│   │   │   ├── context.py               # Smart routing logic
│   │   │   └── memory.py                # Per-window memory management
│   │   ├── input/
│   │   │   ├── monitor/                 # Keystroke monitor package
│   │   │   │   ├── config.py            # Configuration constants
│   │   │   │   ├── state.py             # Global application state
│   │   │   │   ├── threads.py           # Background threads
│   │   │   │   ├── handlers/            # Keystroke & command handlers
│   │   │   │   └── managers/            # IPC, window, log, live-mode managers
│   │   │   ├── monitor.py               # Monitor entry point
│   │   │   ├── injector.py              # Text injection (pynput)
│   │   │   └── typer.py                 # Human-like code typing
│   │   └── media/
│   │       ├── vision.py                # Screenshot + AI vision analysis
│   │       └── voice.py                 # Speech recognition
│   │
│   └── renderer/                        # Electron frontend
│       ├── overlay/
│       │   ├── index.html               # Main overlay UI
│       │   └── index.js                 # Overlay window logic
│       ├── chat/
│       │   ├── index.html               # Persistent chat interface
│       │   └── index.js                 # Chat window logic
│       ├── settings/
│       │   ├── index.html               # Settings UI
│       │   └── index.js                 # Settings logic
│       ├── output/
│       │   └── index.html               # Suggestion popup
│       ├── explanation/
│       │   └── index.html               # Code explanation window
│       ├── common/
│       │   └── styles.css               # Shared styles
│       └── lib/
│           └── marked.min.js            # Markdown parser
│
├── tests/                               # Test suite
├── tools/
│   └── dev/                             # Developer analysis scripts
│       ├── math_analysis.py
│       ├── keystroke_analysis.py
│       └── typer_simulator.py
├── data/                                # Runtime-generated files
│   ├── chats/                           # Persistent chat history
│   └── memory/                          # Per-window conversation memory
└── docs/                                # Documentation

Requirements

• Node.js 16+
• Python 3.8+
• Windows 10/11 (primary support)
• macOS/Linux (partial support)

API Keys

Create .env file with your API keys. Prod_Layer supports 12+ providers. Keys can also be configured through the Settings UI (Ctrl+Shift+S).

All Agents & API Key Links

Fast Models

Agent	Env var	Get API key
Groq Llama 3.3 70B	GROQ_API_KEY	https://console.groq.com/keys
Mistral Small	MISTRAL_API_KEY	https://console.mistral.ai/api-keys/
GPT-4o Mini	OPENAI_API_KEY	https://platform.openai.com/api-keys
Gemini 2.0 Flash	GEMINI_API_KEY	https://aistudio.google.com/apikey

Powerful Models

Agent	Env var	Get API key
DeepSeek Chat	DEEPSEEK_API_KEY	https://platform.deepseek.com/api_keys
Claude Sonnet	ANTHROPIC_API_KEY	https://console.anthropic.com/settings/keys
GPT-4o	OPENAI_API_KEY	https://platform.openai.com/api-keys
Grok 4	XAI_API_KEY	https://console.x.ai/
Together Llama 3.3 70B	TOGETHERAI_API_KEY	https://api.together.xyz/settings/api-keys
Perplexity Sonar Pro	PERPLEXITYAI_API_KEY	https://www.perplexity.ai/settings/api
Cohere Command R+	COHERE_API_KEY	https://dashboard.cohere.com/api-keys
Replicate Llama 405B	REPLICATE_API_TOKEN	https://replicate.com/account/api-tokens

Reasoning Models

Agent	Env var	Get API key
DeepSeek Reasoner	DEEPSEEK_API_KEY	https://platform.deepseek.com/api_keys

Vision (Screenshots)

Provider	Env var	Get API key
Google / Gemini Vision	GOOGLE_API_KEY or GEMINI_API_KEY	https://aistudio.google.com/apikey

Example .env

# Primary providers
MISTRAL_API_KEY=your-mistral-key
GROQ_API_KEY=your-groq-key
DEEPSEEK_API_KEY=your-deepseek-key

# Additional providers (optional)
ANTHROPIC_API_KEY=your-anthropic-key
OPENAI_API_KEY=your-openai-key
GEMINI_API_KEY=your-gemini-key
XAI_API_KEY=your-xai-key
TOGETHERAI_API_KEY=your-togetherai-key
PERPLEXITYAI_API_KEY=your-perplexity-key
COHERE_API_KEY=your-cohere-key
REPLICATE_API_TOKEN=your-replicate-token
GOOGLE_API_KEY=your-google-key

Troubleshooting

Problem	Solution
Shortcuts not working	Run as Administrator
Module not found	pip install -r requirements.txt
Unicode errors	Set PYTHONIOENCODING=utf-8
API errors	Check keys in .env
Window visible in screen share	Restart app after enabling

License

MIT License