With our Prompts management feature, you can easily iterate and version your prompts as a team, deploy them to your codebase with one line of code, share them with your team, test them in the Playground, and more.

Check out the Prompts page to see the prompts you have created and deployed.

​
How to create a prompt

1

Create a prompt

Go to Prompts page, click on the+ New prompt button, and fill in the required fields.

2

Write your prompt

You can write your prompt in the editor and use variables in your prompt by using double curly braces {{variable_name}}.

3

Configure your prompt

You can configure your prompt by specify model, temperature, max_tokens, top_p, frequency_penalty, presence_penalty, and stream.

4

Test your prompt

After you have created your prompt, you can test it in the Playground to see the outputs. Click on the Playground button to test your prompt.

5

Deploy your prompt to the production

After you have tested your prompt and are satisfied with the outputs, you can commit it and then deploy it to your codebase with one line of code.

Once you commit the prompt, you should click on the Commit history button and choose the previous read-only version to deploy it.

​
Example

{
  "prompt": {
     "prompt_id": "prompt_id", //paste this from the prompt management page
     "variables": {
       "variable_name": "variable_value"
     },
     "version": 1 //specify which version of the prompt you want to use
     // "echo": true //optional parameter
  }
}

​
Optional parameter

echo
boolean
default: true

with echo on, the response body will have an extra field

  "prompt_message": [an array of messages]
override
boolean
default: true

Turn on override to use params in override_params instead of the params in the prompt.

{
  "override": true,
}
override_params
boolean
default: true

You can put any OpenAI chat/completions parameters here to override the prompt’s parameters. This will only work if override is set to true.

{
  "override_params": {
    "temperature": 0.5,
    "max_tokens": 100
  }
}
Check out all Keywords AI supported params here.

​
Enable steam when you’re using OpenAI SDK

If you’re using OpenAI SDK and want to connect with the prompt you created, you have to specify stream=True in the call body.

LLM-as-a-judgeLLM playground