Quick Start Guide

Run Prompt Sail on your local machine

Prompt Sail is built as a set of docker containers. One for the backend (promptsail-backend) and one for the frontend (promptsail-ui).

  • promptsail-backend is a proxy that sits between your LLM framework of choice (LangChain, OpenAI python lib etc) and LLM provider API. You change api_base to point to Prompt Sail proxy_url and then it will capture and log all your prompts and responses.
  • promptsail-ui is a user interface that allows you to view, search and analyze all transactions (prompts and responses).

There are two options to run the Prompt Sail docker containers:

Build the Docker images from the source code

Building from a source will give you the latest version of the code with the newest features. However, please note that there might be uncaught bugs that could affect the stability of the application.

Clone the repository from GitHub.

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

To build the Docker images, execute the following command. This will build the images for the backend and UI, and pull the MongoDB and Mongo Express images from Docker Hub. All environment variables are set for you, but you can change them.

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

Pull and run the Docker images from GHCR

Pulling the images from GHCR will give you the latest stable version of the code, however, you will not have the latest features.

The prepared docker-compose file will pull Prompt Sail (backend,ui) images from GitHub Container Registry with latest tags, also it will pull the latest mongo and mongo-express images from Docker Hub:

docker pull ghcr.io/promptsail/promptsail-ui:latest
docker pull ghcr.io/promptsail/promptsail-backend:latest

If you’ve previously pulled Prompt Sail images from ghcr, ensure to pull the promptsail-ui and promptsail-backend images before executing the docker-compose command. This will update any previously pulled images.

docker-compose -f docker-compose.yml up

If you want to run the dev version of the images, you can pull the dev-release tag insted of latest. More on image tagging strategy and deployments can be found at Deployment](/docs/deploy-promptsail-local#pull-and-run-the-docker-images-from-ghcr) Cookbook - Local Deployment](/docs/deploy-promptsail-local#pull-and-run-the-docker-images-from-ghcr) section.

All the environment variables are set to default and non-production deployment in the docker-compose.yml it is recommended to change them to your own values.

Check that the Docker containers are running

🖥️ User Interface (UI)

The UI should be up and running at http://localhost:80/.

  • Default organization name: PromptSail
  • 🛠️ You can edit the organization name in the database using mongo-express.

🔧 Backend Service (API)

The backend services should be operational at http://localhost:8000/.

🗄️ MongoDB

The MongoDB database should be running at http://localhost:27017/.

  • Default login credentials: root:password
  • Default database name: prompt_sail
  • Default folder for storing data will be located in the root directory of the project in the data/mongo folder.

📊 Mongo-Express

Mongo-Express acts as a web-based MongoDB admin interface. It should be accessible at http://localhost:8081/.

  • Default login credentials: admin:pass
  • It is not necessary to use Mongo-Express to run Prompt Sail, but it can help debug and monitor the database.

All the settings can be changed in the appropriate dokcer-compose files:

Create your first project and add at least one AI provider

In the Prompt Sail UI, go to your Organization’s dashboard . Using the Add new project form, create your first project and add at least one AI provider.

Make your first Gen AI API call

OpenAI Chat model example

Based on How to store transactions from OpenAI via OpenAI Python SDK notebook.

Create .env file with your OpenAI API key and organization ID.

OPENAI_API_KEY=sk-xxx
OPENAI_ORG_ID=org-xxx

Import OpenAI class from openai module and create an instance of the class with your OpenAI API key and organization ID.


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")

To make an API call to OpenAI via Prompt Sail, you will need a proxy_url. This can be obtained in the Prompt Sail UI - under the AI Providers tab in your Project’s Dashboard. Before continuing, make sure that the OpenAI is in your project’s AI providers list. If not, you will need to add it first.

Once you have the auto-generated proxy-url, replace the api_base address with it in your code.

More about proxy_url you can learn here


api_base = "http://localhost:8000/projectxyz/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)

It is also possible to tag Gen AI API calls by passing tags in the proxy_url. To simplify this process, use the proxy url generator found in the AI Providers tab of your Project Dashboard. This tool is available for every AI provider you’ve added.

More about the structure of proxy_url (aka api_base) for passing tags here


api_base = "http://localhost:8000/project1/openai/?tags=zero_shot,simple_prompt,dev1,poc&target_path="

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)

More examples

You can find more examples as jupyter notebooks in the repository folder prompt_sail/examples. All examples of tested integrations are also documented in Gen AI API integrations section.

Updated: