docs/getting_started.md

Getting Started with LLMPromptKit

This guide will help you get started with LLMPromptKit, a comprehensive library for managing LLM prompts.

Installation

pip install llmpromptkit

Basic Usage

Initialize Components

from llmpromptkit import PromptManager, VersionControl, PromptTesting, Evaluator

# Initialize with default storage location
prompt_manager = PromptManager()

# Or specify a custom storage location
# prompt_manager = PromptManager("/path/to/storage")

# Initialize other components
version_control = VersionControl(prompt_manager)
testing = PromptTesting(prompt_manager)
evaluator = Evaluator(prompt_manager)

Create and Manage Prompts

# Create a prompt
prompt = prompt_manager.create(
    content="Translate the following text from {source_language} to {target_language}: {text}",
    name="Translation Prompt",
    description="A prompt for translating text between languages",
    tags=["translation", "multilingual"]
)

# The prompt.id property contains a unique identifier (e.g., "a1b2c3d4e5")
prompt_id = prompt.id

# Get a prompt by ID
retrieved_prompt = prompt_manager.get(prompt_id)

# Update a prompt
prompt_manager.update(
    prompt_id,
    content="Please translate the following text from {source_language} to {target_language}:\n\n{text}"
)

# Search prompts by tags
translation_prompts = prompt_manager.search_by_tags(["translation"])

# List all prompts
all_prompts = prompt_manager.list_all()

Version Control

# Create a version snapshot
version_control.commit(
    prompt_id=prompt_id,
    commit_message="Initial version"
)

# Update the prompt and create another version
prompt_manager.update(
    prompt_id,
    content="Please provide a translation of the following text from {source_language} to {target_language}:\n\n{text}\n\nMaintain the original formatting and tone."
)

version_control.commit(
    prompt_id=prompt_id,
    commit_message="Added formatting instructions"
)

# List all versions
versions = version_control.list_versions(prompt_id)

# Compare versions
diff = version_control.diff(prompt_id, 1, 2)

# Revert to a previous version
version_control.checkout(prompt_id, 1)

Using Prompts with Variables

# Get a prompt
prompt = prompt_manager.get(prompt_id)

# Render with variables
rendered_prompt = prompt.render(
    source_language="English",
    target_language="Spanish",
    text="Hello, how are you today?"
)

# Now use rendered_prompt with your LLM API

Next Steps

• See the CLI Usage guide for command-line operations
• Explore Advanced Features for templating and custom metrics
• Check Integration Examples for real-world use cases