Skip to main content

promptic

90% of what you need for LLM app development. Nothing you don't.

promptic is a lightweight, decorator-based Python library that simplifies the process of interacting with large language models (LLMs) using litellm. With promptic, you can effortlessly create prompts, handle input arguments, receive structured outputs from LLMs, and build agents with just a few lines of code.

Installation

pip install promptic

Usage

Basics

from promptic import llm

@llm
def president(year):
    """Who was the President of the United States in {year}?"""

print(president(2000))
# The President of the United States in 2000 was Bill Clinton until January 20th, when George W. Bush was inaugurated as the 43rd President.

Structured Outputs

from pydantic import BaseModel
from promptic import llm

class Capital(BaseModel):
    country: str
    capital: str

@llm
def capital(country) -> Capital:
    """What's the capital of {country}?"""

print(capital("France"))
# country='France' capital='Paris'

Agents

from datetime import datetime

from promptic import llm

@llm(
    system="You are a helpful assistant that manages schedules and reminders",
    model="gpt-4o-mini"
)
def scheduler(command):
    """{command}"""

@scheduler.tool
def get_current_time():
    """Get the current time"""
    print("getting current time")
    return datetime.now().strftime("%I:%M %p")

@scheduler.tool
def add_reminder(task: str, time: str):
    """Add a reminder for a specific task and time"""
    print(f"adding reminder: {task} at {time}")
    return f"Reminder set: {task} at {time}"

@scheduler.tool
def check_calendar(date: str):
    """Check calendar for a specific date"""
    print(f"checking calendar for {date}")
    return f"Calendar checked for {date}: No conflicts found"

cmd = """
What time is it? 
Also, can you check my calendar for tomorrow 
and set a reminder for a team meeting at 2pm?
"""

print(scheduler(cmd))
# getting current time
# checking calendar for 2023-10-05
# adding reminder: Team meeting at 2023-10-05T14:00:00
# The current time is 3:48 PM. I checked your calendar for tomorrow, and there are no conflicts. I've also set a reminder for your team meeting at 2 PM tomorrow.

Streaming

The streaming feature allows real-time response generation, useful for long-form content or interactive applications:

from promptic import llm

@llm(stream=True)
def generate_article(topic):
    """Write a detailed article about {topic}. Include introduction, 
    main points, and conclusion."""

print("".join(generate_article("artificial intelligence")))

Error Handling and Dry Runs

Dry runs allow you to see which tools will be called and their arguments without invoking the decorated tool functions. You can also enable debug mode for more detailed logging.

from promptic import llm

@llm(
    system="you are a posh smart home assistant named Jarvis",
    dry_run=True,
    debug=True,
)
def jarvis(command):
    """{command}"""

@jarvis.tool
def turn_light_on():
    """turn light on"""
    return True

@jarvis.tool
def get_current_weather(location: str, unit: str = "fahrenheit"):
    """Get the current weather in a given location"""
    return f"The weather in {location} is 45 degrees {unit}"

print(jarvis("Please turn the light on and check the weather in San Francisco"))
# ...
# 2024-11-21 13:29:08,587 - promptic - INFO - promptic.py:185 - [DRY RUN]: function_name = 'turn_light_on' function_args = {}
# 2024-11-21 13:29:08,587 - promptic - INFO - promptic.py:185 - [DRY RUN]: function_name = 'get_current_weather' function_args = {'location': 'San Francisco'}
# ...

Resilient LLM Calls with Tenacity

promptic pairs perfectly with tenacity for handling temporary API failures, rate limits, validation errors, and so on. Here's how you can implement a cost-effective retry strategy that starts with smaller models:

from tenacity import retry, stop_after_attempt, retry_if_exception_type
from pydantic import BaseModel, ValidationError
from promptic import llm

class MovieReview(BaseModel):
    title: str
    rating: float
    summary: str
    recommended: bool

@retry(
    # Retry only on Pydantic validation errors
    retry=retry_if_exception_type(ValidationError),
    # Try up to 3 times
    stop=stop_after_attempt(3),
)
@llm(model="gpt-3.5-turbo")  # Start with a faster, cheaper model
def analyze_movie(text) -> MovieReview:
    """Analyze this movie review and extract the key information: {text}"""

try:
    # First attempt with smaller model
    result = analyze_movie("The new Dune movie was spectacular...")
except ValidationError as e:
    # If validation fails after retries with smaller model, 
    # try one final time with a more capable model
    analyze_movie.retry.stop = stop_after_attempt(1)  # Only try once with GPT-4o
    analyze_movie.model = "gpt-4o"
    result = analyze_movie("The new Dune movie was spectacular...")

print(result)
# title='Dune' rating=9.5 summary='A spectacular sci-fi epic...' recommended=True

This approach:

  1. Starts with a faster, cheaper model (gpt-3.5-turbo)
  2. Retries only on Pydantic validation errors
  3. Falls back to a more capable model (gpt-4o) if validation still fails

Features

  • Decorator-based API: Easily define prompts using function docstrings and decorate them with @promptic.llm.
  • Argument interpolation: Automatically interpolate function arguments into the prompt using {argument_name} placeholders within docstrings.
  • Pydantic model support: Specify the expected output structure using Pydantic models, and promptic will ensure the LLM's response conforms to the defined schema.
  • Streaming support: Receive LLM responses in real-time by setting stream=True when calling the decorated function.
  • Simplified LLM interaction: No need to remember the exact shape of the OpenAPI response object or other LLM-specific details. promptic abstracts away the complexities, allowing you to focus on defining prompts and receiving structured outputs.
  • Build Agents Seamlessly: Decorate functions as tools that the LLM can use to perform actions or retrieve information.

Why promptic?

promptic is designed to be simple, functional, and robust, providing exactly what you need 90% of the time when working with LLMs. It eliminates the need to remember the specific shapes of OpenAPI response objects or other LLM-specific details, allowing you to focus on creating prompts and receiving structured outputs.

With its legible and concise codebase, promptic is reliable easy to understand. It leverages the power of litellm under the hood, ensuring compatibility with a wide range of LLMs.

License

promptic is open-source software licensed under the Apache License 2.0.

Metadata

Release files for promptic 1.1.7

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for promptic 1.1.7
File Size Uploaded Provenance Actions
promptic-1.1.7.tar.gz 78.6 kB With provenance Details for promptic-1.1.7.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for promptic 1.1.7
File Interpreter ABI Platform Actions
promptic-1.1.7-py2.py3-none-any.whl Python 3, Python 2 none any Details for promptic-1.1.7-py2.py3-none-any.whl

Total release size: 89.2 kB

Release files / promptic-1.1.7.tar.gz

Download URL promptic-1.1.7.tar.gz
Size 78.6 kB
Tags Source
SHA-256 checksum
How to use checksums
9515d1c958d12d54fd3d3df37c507443eb36d9894746a177732487efe2ce4552
BLAKE2b-256 checksum
How to use checksums
a5af24ffe32df4588418848aa5393551140a048cbd59a942f854c014faa08385
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/5.1.1 CPython/3.12.7

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 21, 2024.

Transparency log

Release files / promptic-1.1.7-py2.py3-none-any.whl

Download URL promptic-1.1.7-py2.py3-none-any.whl
Size 10.6 kB
Tags Python 2 Python 3
SHA-256 checksum
How to use checksums
4f86eef66e9c2ffe7ec2290f7b88d2a361cbfef1d308aaba71c1993d29878658
BLAKE2b-256 checksum
How to use checksums
d6bec9f39b452686f95a590e72d0193054ad908a819c86654af43d600f404261
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/5.1.1 CPython/3.12.7

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 21, 2024.

Transparency log

Security for promptic 1.1.7

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for promptic-1.1.7.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 21, 2024.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

1.1.7 This release

2 release files Full provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page