Prompt Studio

A lightweight, offline-first desktop GUI for managing a CSV library of prompt "roles", composing/running prompts against multiple LLM backends (OpenAI API, Ollama, LM Studio), auto-categorizing prompts, and building runnable drag-and-drop prompt workflows.

Features

• Prompt Manager & Launcher: Import from CSV, browse with categories/tags, search/filter, CRUD operations, favorites, and history
• Smart Prompt Composer: Dynamic placeholders with types and defaults, parameter presets, template preview
• API Connector: Pluggable connectors for OpenAI API, Ollama, LM Studio with streaming support
• Auto-Categorization: LLM-powered prompt analysis and category/tag suggestions
• Workflow Builder: Drag-and-drop node editor for building prompt pipelines (planned)
• Cross-platform: Works on macOS and Windows

Installation

1. Prerequisites: Python 3.11 or higher

2. Install dependencies:

   pip install -r requirements.txt

3. Run the application:

   python main.py

Initial Setup

On first launch, the application will automatically import the sample prompts from assets.csv. The database file prompt_studio.db will be created in the application directory.

Sample Prompts Dataset

The included sample prompts are sourced from the Awesome ChatGPT Prompts dataset by fka on Hugging Face, available at: https://huggingface.co/datasets/fka/awesome-chatgpt-prompts

This dataset is licensed under CC0-1.0 (Creative Commons Zero), making it freely available for any use.

Usage

Importing Prompts

1. Go to File → Import CSV...
2. Select your CSV file with the following expected columns:
   • name (required): The prompt name/title
   • content or prompt (required): The actual prompt text
   • category (optional): Category for organization
   • tags (optional): Comma-separated tags
   • description (optional): Prompt description
   • placeholders_schema (optional): JSON schema for dynamic placeholders

Using Prompts

1. Select a prompt from the center list
2. Fill in parameters in the right panel (if the prompt has placeholders)
3. Choose a backend from the toolbar (OpenAI, Ollama, LM Studio)
4. Select a model for the chosen backend
5. Click Run or press Ctrl+Enter

Keyboard Shortcuts

• Ctrl+K: Quick open (fuzzy search prompts)
• Ctrl+Enter: Run current prompt
• Escape: Stop execution
• F5: Refresh prompt list
• Ctrl+N: New prompt (planned)

Placeholders and Templates

Prompts support Jinja2 templating with dynamic placeholders:

Act as a {{ expertise }} and help me with {{ task }}.
Constraints: {{ constraints or "none" }}

Supported placeholder types:

• str: Short text input
• text: Long text (textarea)
• int: Integer input
• float: Float input
• bool: Boolean checkbox
• choice: Single choice dropdown
• multichoice: Multiple choice checkboxes

Configuration

API Keys

API keys are stored securely using the system keyring:

• OpenAI: Stored under service "PromptStudio", account "openai_api_key"
• Other backends: Local servers (Ollama, LM Studio) don't require API keys

You can set API keys programmatically:

import keyring
keyring.set_password("PromptStudio", "openai_api_key", "your_api_key_here")

Backend Configuration

• OpenAI API: Default endpoint is https://api.openai.com/v1
• Ollama: Default endpoint is http://localhost:11434
• LM Studio: Default endpoint is http://localhost:1234/v1

Settings can be modified through the Settings dialog (planned) or directly in the database.

Development

Project Structure

prompt_studio/
├── models/          # Database models and ORM
├── ui/              # PySide6 user interface components
├── backends/        # LLM backend connectors
├── utils/           # Utilities (CSV import, templating)
└── __init__.py

main.py              # Application entry point
requirements.txt     # Python dependencies
assets.csv          # Sample prompts data

Adding New Backends

To add a new LLM backend:

1. Implement the LLMBackend protocol in backends/llm_backends.py
2. Add your backend to the BackendManager
3. The UI will automatically detect and include it

Building for Distribution

Use PyInstaller to create standalone executables:

# For current platform
pyinstaller --windowed --onefile main.py

# For macOS app bundle
pyinstaller --windowed --onefile --name "Prompt Studio" main.py

TODO / Planned Features

• Settings dialog with API key management
• Quick open fuzzy search dialog
• Export functionality (JSON, CSV)
• Workflow builder with drag-and-drop nodes
• Auto-categorization using LLMs
• History search and filtering
• Dark/light theme toggle
• Plugin system for custom backends
• Prompt templates marketplace
• Batch processing capabilities

License

This project is open source. See LICENSE file for details.

Contributing

Contributions are welcome! Please feel free to submit pull requests or open issues for bugs and feature requests.