philiprehberger-prompt-builder 0.2.1

Type-safe prompt template builder for LLM APIs

philiprehberger-prompt-builder

Type-safe prompt template builder for LLM APIs.

Installation

pip install philiprehberger-prompt-builder

Usage

Fluent Builder

from philiprehberger_prompt_builder import Prompt

messages = (
    Prompt()
    .system("You are a helpful assistant.")
    .user("Summarize this article: {article}")
    .render(article="Long article text here...")
)
# [{"role": "system", "content": "You are a helpful assistant."},
#  {"role": "user", "content": "Summarize this article: Long article text here..."}]

Few-Shot Examples

messages = (
    Prompt()
    .system("Classify the sentiment.")
    .example(user="I love this!", assistant="positive")
    .example(user="Terrible product.", assistant="negative")
    .user("{text}")
    .render(text="Pretty good actually")
)

Conditional Content

use_examples = True

messages = (
    Prompt()
    .system("You are a helpful assistant.")
    .conditional(use_examples, "user", "Here are some examples...")
    .conditional(use_examples, "assistant", "I understand the examples.")
    .user("Now answer my question: {question}")
    .render(question="What is Python?")
)

Prompt Composition

# Build reusable prompt fragments
preamble = Prompt().system("You are a coding assistant.").user("Use Python 3.12+.")
task = Prompt().user("Write a function that {task}")

# Merge fragments into a single prompt
combined = preamble.merge(task)
messages = combined.render(task="sorts a list")

Reusable Templates

from philiprehberger_prompt_builder import PromptTemplate

summarizer = PromptTemplate(
    system="You are a {tone} summarizer. Output in {format}.",
    user="Summarize: {content}",
    defaults={"tone": "concise", "format": "bullet points"},
)

# Use with defaults
messages = summarizer.render(content="Article text...")

# Override defaults
messages = summarizer.render(content="...", tone="detailed", format="paragraphs")

# Create variant
verbose = summarizer.extend(tone="thorough", format="essay")

Token Estimation

prompt = Prompt().system("...").user("{text}")
estimated = prompt.estimate_tokens(text="Hello world")

API

Function / Class	Description
Prompt	Fluent builder for constructing LLM message lists
.system(content)	Add a system message
.user(content)	Add a user message
.assistant(content)	Add an assistant message
.message(role, content)	Add a message with any role
.example(user, assistant)	Add a few-shot example pair
.conditional(include, role, content)	Conditionally add a message if include is truthy
.merge(other)	Create a new Prompt combining messages from self and other
.render(**kwargs)	Render with variable substitution, returns list of dicts
.render_messages(**kwargs)	Render and return Message objects
.estimate_tokens(**kwargs)	Rough token count (~4 chars/token)
PromptTemplate	Reusable prompt template with default values
Message	A single message with role and content

Development

pip install -e .
python -m pytest tests/ -v

Support

If you find this project useful:

⭐ Star the repo

🐛 Report issues

💡 Suggest features

❤️ Sponsor development

🌐 All Open Source Projects

💻 GitHub Profile

🔗 LinkedIn Profile

License

MIT