Skip to content

Latest commit

 

History

History
213 lines (163 loc) · 10 KB

AGENT_DEFINITION_SCHEMA.md

File metadata and controls

213 lines (163 loc) · 10 KB
Raw

Agent Definition Schema Documentation

This document provides a detailed explanation of the structure for agent definitions in the platform. The schema ensures that both the backend and frontend can understand and interact with the agents in a standardized manner.

Schema Version: 1.0.0

** Top-Level Structure**

The agent definition JSON now includes three main sections with their corresponding descriptions:

  • inputDescription: A general description of all inputs required by the agent.
  • processingDescription: An overview of the agent's processing logic.
  • outputDescription: A description of what the agent's outputs represent.

Fields Overview

Field Type Description Required Example
inputDescription string A brief description of the overall inputs the agent requires. Yes "Data required to perform text translation"
inputs array An array of input definitions, described in detail below. Yes See Inputs section below
processingDescription string A brief description of the processing logic of the agent. Yes "Performs text translation using Python"
processing object An object defining the processing details of the agent. Yes See Processing section below
outputDescription string A brief description of the agent's outputs. Yes "Translated text result"
outputs array An array of output definitions, described in detail below. Yes See Outputs section below

Inputs

Definition:

Each input defines the data that the agent requires from the user. Inputs are used to render forms in the frontend and to validate the data before execution in the backend.

Fields:

Field Type Description Required Example
name string Name of the input. Yes "input1"
type string Type of input (string, number, boolean, dropdown, textarea). Yes "string"
description string A brief description of what the input represents. Yes "The first input"
required boolean Specifies whether the input is mandatory. Yes true
validation object Validation rules for the input (depending on input type: string, file, image). No { "minLength": 3, "maxSize": 1000000 }
values object Only applicable for dropdown. Specifies the possible values users can select. No { "option1": "Option 1", "option2": "Option 2" }

Validation Object

The validation object contains different rules depending on the type of input.

For string inputs:

Field Type Description Required Example
minLength integer Minimum length of the string. Must be greater than or equal to 0. No 3
maxLength integer Maximum length of the string. Must be greater than minLength. No 255
regex string Regular expression pattern the string must follow. No "^[A-Za-z0-9]*$"

Examples

String input with validation:

{
  "name": "username",
  "type": "string",
  "description": "User's username",
  "required": true,
  "validation": {
    "minLength": 5,
    "maxLength": 20,
    "regex": "^[a-zA-Z0-9_]*$"
  }
}

Outputs

Definition:

The output section describes what the agent returns after processing. Outputs can be text, files, or other data types. The frontend will render the appropriate output based on this configuration.

Fields:

Field Type Description Required Example
name string The name of the output. Yes "output1"
type string Type of output (string, file, image). Yes "file"
description string A brief description of what the output represents. Yes "Generated image file"

Examples

Text output:

{
  "name": "translatedText",
  "type": "string",
  "description": "Translated text result"
}

Processing

Definition:

This section defines the processing details of the agent, including the platform, version, runtime, and dependencies needed to execute the agent code. It also includes whether the execution is asynchronous.

Fields:

Field Type Description Required Example
type string The type of task (e.g., Image Generation, Text Generation). Yes "Image Generation"
entrypoint string The entrypoint script where the processing starts. Yes "main.py"
requirements object Information about the platform, version, runtime, and dependencies required to execute the agent. Yes { "platform": "python", "version": "3.6", "runtime": "nodejs14.x"}
async boolean Whether the execution is asynchronous. No true

Requirements Object

The requirements object specifies the platform, version, runtime, and libraries required to run the agent. This is essential for configuring the backend environment.

Field Type Description Required Example
platform string The programming language or platform the agent uses (e.g., "python", "nodejs"). Yes "Python"
version string The specific version of the platform required. Yes "3.6"
runtime string The AWS Lambda runtime environment for the agent (e.g., nodejs14.x, python3.8). Yes "nodejs14.x"
libraries array List of required libraries, including their names and versions. No [{"name": "numpy", "version": "1.18.1"}]

Processing Example

{
  "processing": {
    "description": "Image generation using Python",
    "type": "Image Generation",
    "entrypoint": "main.py",
    "requirements": {
      "platform": "python",
      "version": "3.6",
      "runtime": "python3.8",
      "libraries": [
        {
          "name": "numpy",
          "version": "1.18.1"
        }
      ]
    },
    "async": true
  }
}

Complete JSON Example

{
  "inputDescription": "This section defines the data inputs required by the agent.",
  "inputs": [
    {
      "name": "username",
      "type": "string",
      "description": "User's username",
      "required": true,
      "validation": {
        "minLength": 5,
        "maxLength": 20
      }
    }
  ],
  "processingDescription": "This section outlines how the agent processes the input data.",
  "processing": {
    "type": "Image Generation",
    "entrypoint": "main.py",
    "requirements": {
      "platform": "Python",
      "version": "3.6",
      "runtime": "python3.8",
      "libraries": [
        {
          "name": "numpy",
          "version": "1.18.1"
        }
      ]
    },
    "async": false
  },
  "outputDescription": "This section defines the data outputs generated by the agent.",
  "outputs": [
    {
      "name": "translatedText",
      "type": "string",
      "description": "Translated text result"
    }
  ]
}