Integral Configuration

This section shows the Integral configuration object, the fields in it and what values they accept

Integral Config

Reference object for configuring an Integral. We support configurations in both YAML and JSON depending on what you chose when setting up your CLI

TIP

The model you choose combined with instructions and linked Reason Functions directly contribute to your cost Learn to write concie prompts here

json

{
    // Required: true
    // Give a name to your API.
    // This name may be passed to generative model, so make it as
    // relevant and descriptive as possible. Avoid spaces
    "name": "string"

    // Required: true
    // This represents the system instruction that will be passed to the generative model.
    // Outline all the steps you need the model to perform here.
    // Be as detailed yet as concise as possible
    "instruction": "string"

    // Required: true
    // Choose a generative model to run your requests.
    // See the list of supported models for mode info
    "model": "enum"

    // Required: false
    // Use this to specify which format you want the results to be in.
    // This works especially well with OpenAI models.
    // Reason trys to interpolate json response if json_object is set
    // For OpenAI, ensure you repeat the word JSON in the instruction
    "responseFormat": "json_object|text"

    // Required: false
    // Use tools to add functions to the Reason API
    // Functions should be created before they can be linked here
    "tools": [
        {
        // Required: true
        // Name of the function as defined in Reason Functions
        "name": "string",
        // Required: true
        // The type of tool being attached.
        // Only supported value is 'function'
        "type": "function"
        }
    ]
}

yaml

# Required: true
# Give a name to your {{$frontmatter.integral_name}}.
# This name may be passed to generative model, so make it as
# relevant and descriptive as possible. Avoid spaces
name: 'string'

# Required: true
# This represents the system instruction that will be passed to the generative model.
# Outline all the steps you need the model to perform here.
# Be as detailed yet as concise as possible
instruction: 'string'

# Required: true
# Choose a generative model to run your requests.
# See the list of supported models for mode info
model: 'enum'

# Required: false
# Use this to specify which format you want the results to be in.
# This works especially well with OpenAI models.
# Reason trys to interpolate json response if json_object is set
# For OpenAI, ensure you repeat the word JSON in the instruction
responseFormat: 'json_object|text'

# Required: false
# Use tools to add functions to the Reason API
# Functions should be created before they can be linked here
tools:
  # Required: true
  # Name of the function as defined in Reason Functions
  - name: 'string'
    # Required: true
    # The type of tool being attached.
    # Only supported value is 'function'
    type: 'function'