PromptLab v1.0 ud83dude80

A simple web application for prompt engineering and testing prompts with different LLM models.

Features

• Use Jinja2 templates for creating dynamic prompts
• Test prompts with different LLM providers (HuggingFace, OpenRouter)
• Compare responses across different models
• Track history of prompt tests

Getting Started

To install Make in Windows, install Chocolatey:

• Open a Windows Powershell terminal as an admin
• Paste this code as a single line:

• Run this command: choco install make

Installation

make install

This will install all required dependencies including:

• Dash (for the web interface)
• Jinja2 (for prompt templating)
• Other dependencies

Running the application

make run

The app will be available at http://127.0.0.1:8050/

Setting up API keys

Edit src/config/.env and add your API keys:

HUGGINGFACE_API_KEY = "your-huggingface-key"
HUGGINGFACE_API_URL = "https://api-inference.huggingface.co/models/YourModelName"
OPENROUTER_API_KEY = "your-openrouter-key"
OPENROUTER_API_URL = "https://openrouter.ai/api/v1/chat/completions"

Creating Templates

Prompt templates are stored in the prompts directory as .jinja2 files. Templates use Jinja2 syntax for variable substitution, for example:

Hello {{ name }}, welcome to our {{ product }} service that offers {{ feature }}.

Running Tests

make test

Formatting Code

make format