promptkit-core 0.1.0

Structured Prompt Engineering for LLM Apps

PromptKit

A production-grade library for structured prompt engineering for LLMs. Define, validate, and execute LLM prompts using YAML files with input validation, engine abstraction, and CLI support.

Features

• 📝 YAML-based prompt definitions with Jinja2 templating
• 🔍 Input validation using Pydantic schemas
• 🏗️ Engine abstraction supporting OpenAI and local models
• 💰 Token estimation and cost calculation
• 🖥️ CLI interface for quick prompt execution
• 🧪 Fully tested with comprehensive test suite

Installation

pip install promptkit-core

For development:

pip install promptkit-core[dev]

📚 Documentation

For detailed usage and examples, please refer to the PromptKit Documentation.

Repository

Source code: https://github.com/ochotzas/promptkit

Quick Start

1. Define a prompt in YAML

Create greet_user.yaml:

name: greet_user
description: Basic greeting
template: |
  Hello {{ name }}, how can I help you today?
input_schema:
  name: str

2. Use in Python

from promptkit.core.loader import load_prompt
from promptkit.core.runner import run_prompt
from promptkit.engines.openai import OpenAIEngine

# Load prompt from YAML
prompt = load_prompt("greet_user")

# Configure engine
engine = OpenAIEngine(api_key="sk-...")

# Run prompt
response = run_prompt(prompt, {"name": "Alice"}, engine)
print(response)

3. Use the CLI

# Run a prompt
promptkit run greet_user --key sk-... --name Alice

# Just render the template
promptkit render greet_user --name Alice

# Validate prompt structure
promptkit lint greet_user

Prompt YAML Structure

name: prompt_identifier
description: What this prompt does
template: |
  Your Jinja2 template here with {{ variables }}
input_schema:
  variable_name: str
  another_var: int
  optional_var: "str | None"

Supported Engines

• OpenAI: Complete OpenAI API support
• Ollama: Local model support (work in progress)

Token Estimation

from promptkit.utils.tokens import estimate_tokens, estimate_cost

tokens = estimate_tokens("Your prompt text here")
cost = estimate_cost(tokens, "gpt-4")
print(f"Estimated cost: ${cost:.4f}")

Development

# Clone the repository
git clone https://github.com/ochotzas/promptkit.git
cd promptkit

# Install development dependencies
pip install -e .[dev]

# Run tests
pytest

# Format code
black .
isort .

# Type checking
mypy promptkit/

License

MIT License - see LICENSE file for details.