Page cover

Introduction

Introduction

Welcome to the Agent Orbit Python SDK! This SDK offers a streamlined interface to interact with the Agent Orbit REST API, enabling you to create, execute, and manage agents effortlessly.


Usage

Installation

You can install the SDK using pip:

bashCopy codepip install agentorbit

Example Code

After installation, use the following code to interact with the Agent Orbit API:

pythonCopy codeimport agentorbit

# Set your API key
agentorbit.api_key = "API_ORBIT_123456789"  # Replace with your actual API key

# Retrieve an agent by ID
agent = agentorbit.Agent.retrieve(id="abc12345-xyz67890-test98765")

# Execute the agent
execution = agent.execute()

Resources Overview

The following resources provide a fundamental understanding of Agent Orbit SDK capabilities:


1. BuilderJob

  • Purpose: Automatically created whenever an Agent is created or when its python_version or dependencies attribute is updated.

  • Output: Produces a Build Instance.

  • Note: BuilderJob and Build Instances cannot be manually created. They are automatically generated by Agent Orbit and available only for listing or retrieval.


2. ExecutorJob

  • Purpose: Executes an Agent for a single task.

  • Output: Produces an Execution Instance.

  • Creation: Automatically initiated in the following cases:

    • When an Agent is executed via the Agent Orbit UI.

    • When the execute method is invoked in Python:

pythonCopy codeexecution = agent.execute()

3. Agent

  • Purpose: The central entity in Agent Orbit, representing programmable logic that can be executed via an ExecutorJob.


4. Build Instance

  • Purpose: Represents a single successful build of an Agent.

  • Creation: Automatically generated by a BuilderJob.


5. Execution Instance

  • Purpose: Represents a single task execution by an Agent.

  • Creation: Generated by an ExecutorJob.


6. DataStore

  • Purpose: A key-value database that can be associated with multiple Agents during their creation.

  • Usage: Accessed via the datastore argument in the main function of an Agent.

  • Example:

pythonCopy codedef main(request, datastore):
    # Set a value in the datastore
    datastore["username"] = "AgentOrbitUser"

    # Retrieve the value
    return datastore["username"]

Conceptual API Details

Here’s an example of how the API endpoints might look for Agent Orbit:

  • Base URL: https://api.agentorbit.ai/v1

  • Endpoints:

    • Retrieve Agent: GET /agents/{agent_id}

    • Execute Agent: POST /agents/{agent_id}/execute

    • DataStore Access: GET /datastore/{store_id}

  • Headers:

    jsonCopy code{
        "Authorization": "Bearer API_ORBIT_123456789",
        "Content-Type": "application/json"
    }

This conceptual display illustrates how Agent Orbit could be structured for users to integrate seamlessly with the API. Even though this is just a concept, the design emulates a realistic and intuitive interface for developers. Let me know if further refinements are needed!

Last updated