AIPromptBridge

• AI Desktop Tools & Integration Bridge •

AIPromptBridge is a Windows desktop application that brings AI assistance to your fingertips. Use global hotkeys to edit text using AI, capture and analyze screen content, and chat with models—all from a lightweight system tray app.

📽️ Demo

Demo-Walkthrough.mp4

✨ Features

🎯 TextEditTool

Press Ctrl+Space anywhere to invoke AI on selected text:

• Understand - Explain, Generate Summaries, or Keypoints
• Edit - Proofread (✏️), Rewrite (📝), or make it Casual (😎)
• Q&A - Use the second input box in the popup to ask any question about the text
• Custom prompts - Define and group your own actions in the Prompt Editor

Works in any application: browsers, IDEs, Notepad, Word, everywhere.

📸 Screen Snip (SnipTool)

Press Ctrl+Shift+X to capture a region of your screen and analyze it with AI:

• OCR - Extract text formatting and structure
• Analysis - Describe, Summarize, or Explain Code
• Data - Extract Data to tables or Transcribe handwriting
• Compare - Compare Images to analyze differences between two screenshots
• Chat - Ask follow-up questions about the captured image

💬 Chat Interface

Lightweight chat windows with:

• Streaming responses (real-time typing)
• Markdown rendering
• Session history (browse and restore)
• Multi-theme UI with 7 color schemes

🎨 Theme System

Customizable appearance with:

• 7 themes: Catppuccin, Dracula, Nord, Gruvbox, OneDark, Minimal, High Contrast
• Dark/Light modes: Each theme has both variants
• System detection: Auto-switches based on Windows theme
• Live preview: See theme changes instantly in Settings

🔄 Robust Backend

• Multi-provider support - Google Gemini, OpenRouter, custom endpoints
• Automatic key rotation - Switch API keys on rate limits (429, 401, 403)
• Smart retry logic - Handles errors gracefully with configurable delays
• Empty response detection - Automatically retries with next key
• Streaming support - Real-time responses
• Batch Processing - Async processing for large workloads (Gemini Batch API)
• Attachment Manager - Efficient external storage for session images and files

🧰 Tools System

The File Processor tool enables bulk operations:

• Batch Processing: Process folders of Images, Audio, Code, Text, or PDFs
• Audio Optimization: Reduce file size (mono, sample rate) for efficient AI processing
• Configurable: On-demand tools_config.json creation
• Smart Handling:
  • Large Files: Auto-switches to Gemini Files API or Chunking logic
  • Checkpoints: Resume interrupted jobs or retry failures
  • Interactive Mode: Pause (P), Stop (S), or Abort (Esc) during processing

🚀 Quick Start

Download (Recommended)

1. Download AIPromptBridge.exe from GitHub Releases
2. Run it - on first launch, it creates config.ini and automatically opens the Settings window
3. Enter your API keys in the API Keys tab and click Save
4. The app starts minimized to system tray

From Source (Alternative)

git clone https://github.com/zaxx-q/AIPromptBridge.git
cd AIPromptBridge
pip install -r requirements.txt
python main.py

Configuration

You can configure API keys via the Settings window (right-click tray icon -> Settings) or by editing config.ini.

[google]
AIzaSyXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX  # My Primary Key

[openrouter]
sk-or-v1-XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX # Project Backup

💡 Tip: Add multiple keys (one per line) for automatic rotation when rate limits are hit. You can name your keys using inline comments (#).

📋 Usage

System Tray

Right-click the tray icon for:

• Show/Hide Console - Toggle console visibility
• Session Browser - View chat history
• Settings - Open GUI settings editor
• Prompt Editor - Edit TextEditTool prompts
• Edit config.ini - Open configuration file
• Restart - Restart the application
• Quit - Exit completely

TextEditTool

1. Select text in any application
2. Press Ctrl+Space
3. Choose an action (Proofread, Rewrite, etc.)
4. Text is replaced or opened in chat

Without selection: Opens a quick input bar for direct questions.

SnipTool (Screen Snipping)

1. Press Ctrl+Shift+X
2. Click and drag to select a screen region
3. Choose an action (Describe, Extract Text, etc.) or ask a question
4. Results open in a chat window with the image attached

API Endpoints

Access AI via HTTP POST (Advanced). Endpoints are disabled by default (flask_endpoints_enabled = false) but can be enabled in config.ini for integrations like ShareX.

# Basic OCR
curl -X POST -F "image=@screenshot.png" http://127.0.0.1:5000/ocr

# With chat window
curl -X POST -F "image=@screenshot.png" "http://127.0.0.1:5000/describe?show=yes"

See ShareX Setup Guide for full endpoint documentation.

Console Commands

When console is visible, press these keys:

Key	Action
S	Show system status
O	Open session browser
M	List available models (Use ?N for details, e.g., ?1)
P	Switch AI provider
T	Toggle thinking mode
R	Toggle streaming mode
H	Show help

⚙️ Configuration

Providers

Set your preferred provider in config.ini:

[config]
default_provider = google
google_model = gemini-2.5-flash

Available providers:

• google - Native Gemini API (recommended)
• openrouter - OpenRouter.ai models
• custom - Any OpenAI-compatible endpoint

Custom Endpoints

Add your own endpoints in config.ini:

[endpoints]
# Simple custom endpoint
my_analyzer = Analyze this image and list all objects found.

# Dynamic language endpoint using {lang} placeholder
my_translator = Translate to {lang}. Keep formatting.

Access via http://127.0.0.1:5000/my_analyzer or http://127.0.0.1:5000/my_translator?lang=French

Unified Prompts

Customize all prompts (TextEditTool, SnipTool, and Endpoints) in prompts.json. This file is automatically created from defaults if missing.

Structure

• text_edit_tool: Text manipulation prompts (Ctrl+Space)
• snip_tool: Image analysis prompts (Ctrl+Shift+X)
• endpoints: Flask API endpoint prompts
• _global_settings: Shared settings and modifiers

{
  "text_edit_tool": {
    "Proofread": {
      "icon": "✏",
      "prompt_type": "edit",
      "system_prompt": "You are a meticulous proofreader...",
      "task": "Proofread the following text...",
      "show_chat_window_instead_of_replace": false
    }
  },
  "snip_tool": {
    "Describe": {
      "icon": "🖼️",
      "system_prompt": "You are an image analysis expert...",
      "task": "Describe this image in detail...",
      "show_chat_window": true
    }
  }
}

"icon": "💡",
"prompt_type": "general",
"system_prompt": "You are a knowledgeable teacher...",
"task": "Explain the following text...",
"show_chat_window_instead_of_replace": true

}, "Rewrite": { "icon": "📝", "prompt_type": "edit", "system_prompt": "You are an expert editor...", "task": "Rewrite this text to improve clarity...", "show_chat_window_instead_of_replace": false }, }

### Text Modifiers

The TextEditTool popup includes a **Modifier Bar** that lets you fine-tune the output. Toggle these modifiers to inject specific instructions into the prompt:

- **Variations** (🔢): Generate 3 alternative versions.
- **Direct** (🎯): Make output direct and concise, no fluff.
- **Explain** (📝): Add an explanation of changes.
- **Creative** (🎨): Take more liberties with phrasing.
- **Literal** (📏): Stay close to the original.
- **Shorter** (✂️): Make the result more concise.
- **Longer** (📖): Expand with more detail.
- **Formal** (💼): Professional/business tone.
- **Informal** (💬): Casual/personal tone.
- **Global** (🌐): Avoid idioms for international audience.

*Note: Some modifiers (like Variations and Explain) force the output to open in a chat window.*

### Theming

Configure the UI theme in Settings or `config.ini`:

```ini
[config]
ui_theme = catppuccin
ui_theme_mode = auto  # auto, dark, light

Available themes: catppuccin, dracula, nord, gruvbox, onedark, minimal, highcontrast

💡 Performance Tip: If you experience lag or UI issues with the modern interface, you can disable it by enabling "Force Standard Tkinter" in the Theme tab of Settings. This switches the app to a high-performance fallback mode using standard Windows widgets.

💡 Tips

For Faster Responses

• Use non-reasoning models (e.g., gemini-2.0-flash instead of gemini-2.5-pro)
• Disable thinking mode: Press T in console or set thinking_enabled = false
• Keep streaming enabled for perceived faster responses

For Better Results

• Enable thinking mode for complex tasks
• Use specific prompts in TextEditTool
• Add context when asking questions

API Key Management

• Add multiple API keys (one per line) for automatic rotation
• If one key hits rate limits, the next one is used automatically
• The system tracks exhausted keys and skips them
• Keys rotate on: 429 (rate limit), 401/402/403 (auth errors), empty responses

🔧 Command Line Options

AIPromptBridge.exe                    # Normal start (tray mode, console hidden)
AIPromptBridge.exe --no-tray          # No tray icon, console stays visible
AIPromptBridge.exe --show-console     # Tray mode but keep console visible
AIPromptBridge.exe --no-wt            # Skip Windows Terminal detection

💡 Console View: For the best console experience (including full color emoji support), it is highly recommended to use Windows Terminal. AIPromptBridge will attempt to automatically relaunch in Windows Terminal if detected.

📖 Documentation

• Project Structure - File organization
• Architecture - Technical details
• ShareX Setup - Screenshot integration

🗺️ Roadmap

• Prompt Editor - GUI for editing text_edit_tool_options.json (includes Playground)
• Settings Window - GUI for editing config.ini
• Theme System - Multiple color schemes with dark/light modes
• Colored Emoji - Twemoji-based color emoji rendering in chat and UI widgets
• Localization - Multi-language support for UI
• Modern UI - Migrated to CustomTkinter for modern UI, rounded corners, and other GUI improvements

📝 Requirements

• Windows 10/11 (uses Windows-specific APIs for tray and console)
• Windows Terminal (Highly recommended for better console view and colors)
• Python 3.14+ (if running from source)
• API keys for at least one provider (Google Gemini recommended)

📄 License

MIT License

Attribution & Third-Party Licenses

This project uses Twemoji graphics, licensed under CC-BY 4.0.