promptsail / prompt_sail Goto Github PK

LLM’s proxy for prompt and response governance, monitoring, and analysis. 📊🔍

⚠️ Prompt Sail is currently in Development: Expect breaking changes and bugs! Feedback and contributions are welcome. Please see the Contributing Guide for more information.

Prompt Sail is a proxy for Large Language Models (LLMs) API's such as OpenAI GPT models, Azure OpenAI, Anthropic Clude etc. that allows you to record prompts and responses, analyze costs, generation speed, compare and track trends and changes across various models and projects over time.

To learn more about Prompt Sail’s features and capabilities, see

• Documentation 📖

  • Quick Start Guide - How to run Prompt Sail locally via docker.
  • Advanced Features - API access, transctions capturing logic, and more.
  • Deployment Cookbook - How to deploy Prompt Sail on various cloud providers.
  • Developers Guide - Development environment setup, database schema, testing, and more.

• Examples 💻

  • OpenAI SDK -> OpenAI API
  • Langchain SDK -> OpenAI API
  • OpenAI SDK -> Azure OpenAI
  • Langchain SDK -> Azure OpenAI
  • Langchain SDK -> Azure OpenAI Ada embeddings

• API Reference. -->

The easiest way is to test our demo at https://try-promptsail.azurewebsites.net/ (every new deployment erases the database)

Check out the documentation how to run PromptSail locally via docker.

To try out Start Prompt on your own machine, we recommend using docker-compose.

• installed docker and docker-compose on your machine Windows | Mac | Linux
• git clone repository and navigate to main directory

git clone https://github.com/PromptSail/prompt_sail.git
cd prompt_sail

Build and run the docker image:

docker-compose -f docker-compose-build.yml up --build

Pull and run the Docker images from GHCR:

docker-compose -f docker-compose.yml up

Navigate to http://localhost:80 and add you AI provider of choice.

To use Prompt Sail with openai Python library, you need to set OPENAI_API_BASE environment variable, or modify openai.api_base parameter to point to your Prompt Sail project.

from openai import OpenAI
import os
from dotenv import load_dotenv
from pprint import pprint

load_dotenv()

openai_key = os.getenv("OPENAI_API_KEY")
openai_org_id = os.getenv("OPENAI_ORG_ID")

api_base = "http://localhost:8000/project1/openai/"

ps_client = OpenAI(
    base_url=api_base,
    api_key=openai_key,
)

response = ps_client.chat.completions.create(
    model="gpt-3.5-turbo",
    messages=[
        {
            "role": "system",
            "content": "You are a poetic assistant, skilled in explaining complex programming concepts with creative flair.",
        },
        {
            "role": "user",
            "content": "Compose a poem that explains the concept of recursion in programming.",
        },
    ],
)

pprint(response.choices[0].message)

Using Prompt Sail with langchain is similar:

from langchain.chat_models import ChatOpenAI
from langchain.prompts.chat import (
    ChatPromptTemplate,
    HumanMessagePromptTemplate,
    SystemMessagePromptTemplate,
)
from langchain.schema import HumanMessage, SystemMessage

haiku_prompt = [
    SystemMessage(
        content="You are a poetic assistant, skilled in explaining complex programming concepts with creative flair.",
    ),
    HumanMessage(
        content="Compose a haiku that explains the concept of recursion in programming.",
    ),
]
chat = ChatOpenAI(
    temperature=0.9,
    openai_api_key=openai_key,
    openai_organization=openai_org_id,
    model="gpt-3.5-turbo-1106",
)

chat(haiku_prompt)

• Bugs & requests: file a GitHub ticket 🐞
• For business inquiries: email [email protected]. 📧
• Our website: https://promptsail.com 🌐

Prompt Sail is free and open source, under the MIT license.