Athina’s prompt management system allows you to run prompts on the platform.

​
Run a Prompt on Athina Playground

To run a prompt:

  1. Open https://app.athina.ai.ai/prompt and select the prompt you want to run.
  2. Enter the input variables in the editor.
  3. Select the model you want to run the prompt on, and configure any parameters.
  4. Click Run to generate the output.

​
Run a Prompt Programmatically

You can run a prompt via the API by calling the Run Prompt endpoint with the prompt_slug and input variables.

import os
from athina_client.prompt import Prompt, Slug
from athina_client.keys import AthinaApiKey

AthinaApiKey.set_key(os.getenv('ATHINA_API_KEY'))

Prompt.run(
    slug='test-staging',
    # the following fields are optional
    version=2,
    model="gpt-4o",
    variables={
        "company": "nvidia"
    },
    parameters={
        "temperature": 1,
        "max_tokens": 1000
    },
)
  • variables: Input variables for the prompt.
  • version (optional): The version of the prompt to run. If not specified, the default version will be used.
  • model (optional): The model to run the prompt on. If not specified, we will use the model provided in your commit.
  • parameters (optional): Additional parameters for the model like temperature, max_tokens, etc. If not specified, the default parameters will be used.

Prompts run via the API will be displayed in the Observe page: https://app.athina.ai/observe

​
Get Prompt via API & Run it Yourself

Alternatively, you can store a prompt in Athina and run it yourself. This example demonstrates the same using OpenAI’s API in Python.

import os
from athina_client.prompt import Prompt
from athina_client.keys import AthinaApiKey
from dotenv import load_dotenv
import openai

# Load environment variables

load_dotenv()

# Set up Athina and OpenAI API keys

AthinaApiKey.set_key(os.getenv('ATHINA_API_KEY'))
openai.api_key = os.getenv('OPENAI_API_KEY')

# Get the prompt template from Athina

prompt_template = Prompt.get_default('my-prompt') # Replace 'my-prompt' with your prompt slug

# Extract the messages and parameters

messages = prompt_template.prompt # This gets the messages array
model = prompt_template.model if prompt_template.model else "gpt-4" # Default to gpt-4 if not specified
parameters = prompt_template.parameters or {}

# Make the OpenAI API call

response = openai.chat.completions.create(
model=model,
messages=messages,
temperature=parameters.get('temperature', 1),
max_tokens=parameters.get('max_tokens', 1000),
presence_penalty=parameters.get('presence_penalty', 0),
frequency_penalty=parameters.get('frequency_penalty', 0),
top_p=parameters.get('top_p', 1)
)

# Print the response

print(response.choices[0].message.content)