PromptAssist

Installation: https://github.com/EricJujianZou/PromptAssist/releases/tag/v1.0.0

What is PromptAssist?

PromptAssist is a Windows utility that runs in your system tray with two core abilities:

For the commands below, you can type it in ANY textbox (e.g. ChatGPT, job apps, search bar)

• Snippet Expansion: Type a command (like ::emailstarter) and press SPACE right after to replace it with longer text. Use for code boilerplate, common replies, or text you type often.
• LLM Augmentation: For example, type ::Prompt(your-prompt-here) and press SPACE right after in ChatGPT to transform your prompt with added accuracy and context.

How to Use

Snippets

• Double-click the tray icon to open the dashboard and manage snippets.
• Type a snippet command in ANY text box (e.g., ::emailstarter) and press Space to expand it.

LLM Prompt Augmentation

• Type ::Prompt(your-prompt-here) + space at the end to augment the prompt.
• You will here a sound indicating the prompt is being augmented, and another once it's done.
• Press Ctrl+V to paste the result, or look in your clipboard history by WINDOWS key + v

Quick 30-second video demo:

---

Key Features

• System-Wide: Works in code editors, browsers, and text fields.
• Lightweight: Runs in system tray with minimal resources.
• Customizable: Add, edit, and manage snippets through a dashboard.
• History Tracking: View all your past LLM-augmented prompts with full details.
• Secure: Connects to self-hosted API backend for LLM prompts.
• No Installation: Download and run the .exe file.

Getting Started

1. Download PromptAssist.exe from the release page
2. Run the file
3. The icon appears in your system tray (main laptop screen, task bar, the "^" symbol. You won't find it on a monitor).

History Tab

The History tab in the dashboard lets you:

• View all past prompts: See every query you've sent and the augmented results, sorted by date.
• Double-click entries: Open a detailed view with full text and copy buttons.
• Right-click actions: Copy results, save as snippets, or delete individual entries.
• Toolbar buttons: Quick access to copy, save as snippet, and delete actions.
• Hover tooltips: Preview truncated text before opening the full view.

For Developers: Running from Source

The project has two parts:

1. Backend: FastAPI server connecting to the language model
2. Client: PySide6 GUI running in system tray

Prerequisites

• Python 3.9+
• Git
• Docker Desktop

Step 1: Clone the Repository

git clone https://github.com/EricJujianZou/PromptAssist.git
cd PromptAssist

Step 2: Set Up Redis

1. Start Docker Desktop

2. Run Redis container:

   docker run -d -p 6379:6379 --name prompt-redis redis

Step 3: Set Up the Backend

1. Navigate to backend:

   cd backend_api

2. Create and activate virtual environment:

   python -m venv venv
   .\venv\Scripts\activate  # Windows
   or if that doesn't work:
   .\venv\bin\activate.ps1

3. Install dependencies:

   pip install -r requirements_backend.txt

4. Configure environment:

   • Create .env in backend_api
   • Copy contents from .env.example
   • Set values:
     • VERTEX_AI_PROJECT: Your Google Cloud Project ID
     • VERTEX_AI_LOCATION: Region (e.g., us-central1)
     • BACKEND_API_KEY: Create a random secret key
     • REDIS_URL: Use redis://localhost

5. Authenticate with Google Cloud:

   gcloud auth application-default login

6. Run the backend:

   uvicorn main:app --reload

   Server runs on http://127.0.0.1:8000. Leave this terminal open.

Step 4: Set Up the Client

1. Open a new terminal in project root

2. Create virtual environment:

   python -m venv venv_client
   .\venv_client\Scripts\activate  # Windows

3. Install dependencies:

   pip install -r documentation/requirements.txt

4. Configure environment:

   • Create .env in src
   • Copy from src/.env.example
   • Set values:
     • BACKEND_API_URL: http://127.0.0.1:8000
     • BACKEND_API_KEY: Same key from backend's .env

5. Run the client:

   python run.py

The icon appears in your system tray, connected to your local backend and Redis.

License

MIT License. See LICENSE file.