Orbit Logo
Playground

Playground Endpoint Samples

These are sample endpoints you can create in the Orbit Playground to test different AI agents, SDKs, and configurations. Each sample demonstrates specific capabilities and use cases.

Sample Categories

Strands SDK Agents

Custom Python agents using AWS Strands Agent SDK with specialized business logic.

AI SDK Agents

Vercel AI SDK powered agents with direct provider integration and streaming.

OpenAI SDK Agents

Simple OpenAI SDK integration for direct GPT model access.

Strands SDK Agents

These agents use the AWS Strands Agent SDK with Python backends, providing specialized business functionality.

01 SaaS Agent Sample

Endpoint Configuration:

  • Name: 01 SaaS Agent Sample
  • Type: Agent (Strands SDK)
  • Base URL: http://saas-agent:8000
  • Environment: Sandbox

Description: Sample SaaS agent demonstrating gym/club management functionality with comprehensive member management, class scheduling, and facility operations.

Default Configuration Details:

  • Chat Path: /stream
  • Model: Claude Sonnet 4
  • System Instructions: "You are a helpful gym management assistant. Help users with member management, class scheduling, and facility operations."
  • Memory: Disabled
  • Authentication: None required

Sample Queries:

- "Find premium members who joined in the last month"
- "Sign up a new member named Sarah Jacobs with email sarah1@test.com at club_central with a premium plan."
- "Show the top clubs based on renewals over the past 7 days"

02 Weather Forecast Agent

Endpoint Configuration:

  • Name: 02 Weather Forecast Agent
  • Type: Agent (Strands SDK)
  • Base URL: http://weather-agent:8001
  • Environment: Sandbox

Description: Weather information agent with National Weather Service integration, providing accurate forecasts and weather explanations with structured UI components.

Default Configuration Details:

  • Chat Path: /stream-with-components
  • Model: Claude Sonnet 4
  • System Instructions: "You are a weather information assistant. Provide accurate, helpful weather forecasts and explanations."
  • Memory: Disabled
  • Special Features: Chart.js integration for weather visualization

Sample Queries:

- "What's the weather like in San Francisco?"
- "Give me a 7-day forecast for New York"
- "Show weather trends for this week"

03 RAG Knowledge Agent

Endpoint Configuration:

  • Name: 03 RAG Knowledge Agent
  • Type: Agent (Strands SDK)
  • Base URL: http://rag-kb-agent:8002
  • Environment: Sandbox

Description: RAG-enabled agent with Orbit Framework knowledge base integration, providing accurate, cited information from document sources.

Default Configuration Details:

  • Chat Path: /stream
  • Model: Claude Sonnet 4
  • System Instructions: "You are a knowledge assistant with access to Orbit Framework document sources. Use your knowledge base to provide accurate, cited information."
  • Memory: Disabled
  • Knowledge Base: Orbit Framework documentation

Sample Queries:

- "What is the Orbit Framework?"
- "How do I configure Agents?"
- "Explain Orbit agent architecture"

04 Web Scraper Agent

Endpoint Configuration:

  • Name: 04 Web Scraper Agent
  • Type: Agent (Strands SDK)
  • Base URL: http://scraper-agent:8003
  • Environment: Sandbox

Description: Ethical web scraping agent with structured data extraction capabilities, following robots.txt compliance and responsible scraping practices.

Default Configuration Details:

  • Chat Path: /stream
  • Model: Claude Sonnet 4
  • System Instructions: "You are a web scraping assistant. Help users extract information from websites ethically and responsibly."
  • Memory: Disabled
  • Features: Robots.txt compliance, rate limiting, ethical guidelines

Sample Queries:

- "What information is available on https://httpbin.org/html?"
- "Get news headlines from https://cnn.com"
- "Check: https://thegracelands.club How many guests can the Grand Manor accommodate?"

AI SDK Agents

These agents use the Vercel AI SDK with TypeScript, connecting directly to AI provider APIs.

Agent with OpenAI GPT-5-Mini

Endpoint Configuration:

  • Name: Agent with OpenAI GPT-5-Mini
  • Type: Agent (AI SDK)
  • Base URL: https://api.openai.com/v1
  • Environment: Sandbox

Description: AI SDK agent with Vercel AI framework and OpenAI environment credentials, providing streaming conversational AI.

Default Configuration Details:

  • Chat Path: /v1/chat/completions
  • Model: GPT-5-Mini
  • Provider: OpenAI
  • System Instructions: "You are a helpful AI assistant powered by the AI SDK framework. Provide clear and accurate responses."
  • Authentication: Environment variable (OPENAI_API_KEY)
  • Features: Streaming responses, Vercel AI SDK integration

How to Test:

Ensure OPENAI_API_KEY environment variable is set
Create endpoint with AI SDK configuration
Test general conversational queries
Observe streaming response behavior

Agent with Anthropic Claude

Endpoint Configuration:

  • Name: Agent with Anthropic Claude
  • Type: Agent (AI SDK)
  • Base URL: https://api.anthropic.com
  • Environment: Sandbox

Description: AI SDK agent with Anthropic Claude and environment credentials, leveraging Claude's helpful, harmless, and honest approach.

Default Configuration Details:

  • Chat Path: /v1/messages
  • Model: Claude Sonnet 4
  • Provider: Anthropic
  • System Instructions: "You are Claude, an AI assistant created by Anthropic. Be helpful, harmless, and honest."
  • Authentication: Environment variable (ANTHROPIC_API_KEY)
  • Features: Streaming responses, max tokens configuration

How to Test:

Ensure ANTHROPIC_API_KEY environment variable is set
Create endpoint with AI SDK configuration
Test complex reasoning and analysis queries
Compare response style with other providers

Agent with Google Gemini

Endpoint Configuration:

  • Name: Agent with Google Gemini
  • Type: Agent (AI SDK)
  • Base URL: https://generativelanguage.googleapis.com
  • Environment: Sandbox

Description: AI SDK agent with Google Gemini and environment credentials, showcasing Google's multimodal AI capabilities.

Default Configuration Details:

  • Chat Path: /v1/messages
  • Model: Gemini 2.5 Flash
  • Provider: Google
  • System Instructions: "You are Gemini, a helpful AI assistant created by Google. Provide clear and accurate responses."
  • Authentication: Environment variable (GOOGLE_API_KEY)
  • Features: Temperature control, max output tokens

How to Test:

Ensure GOOGLE_API_KEY environment variable is set
Create endpoint with AI SDK configuration
Test creative and analytical tasks
Evaluate response quality and speed

OpenAI SDK Agents

Simple OpenAI SDK integration for direct access to GPT models.

OpenAI API Agent

Endpoint Configuration:

  • Name: OpenAI API Agent
  • Type: Agent (OpenAI SDK)
  • Base URL: https://api.openai.com
  • Environment: Sandbox

Description: Direct OpenAI GPT-5-Mini conversational agent with environment credentials, providing the simplest integration path.

Default Configuration Details:

  • Chat Path: /chat/completions
  • Model: GPT-5-Mini
  • Provider: OpenAI
  • System Instructions: "You are a helpful AI assistant. Respond naturally and conversationally."
  • Authentication: Environment variable (OPENAI_API_KEY)
  • Features: Streaming enabled, minimal configuration

How to Test:

Ensure OPENAI_API_KEY environment variable is set
Create endpoint with OpenAI SDK configuration
Test with basic conversational queries
Verify simple, direct responses

General Testing Guidelines

Environment Variables Required:

  • OPENAI_API_KEY - For OpenAI-powered endpoints
  • ANTHROPIC_API_KEY - For Anthropic Claude endpoints
  • GOOGLE_API_KEY - For Google Gemini endpoints
  • AWS credentials - For Strands SDK endpoints (will use Orbit's by default)

Testing Best Practices

  1. Start Simple - Begin with basic queries to verify endpoint connectivity
  2. Test SDK Features - Explore SDK-specific capabilities (streaming, tools, components)
  3. Compare Responses - Test similar queries across different endpoints
  4. Monitor Performance - Observe response times and streaming behavior
  5. Debug Issues - Use the Debug tab for Strands SDK endpoints

Common Testing Scenarios

Conversational Testing:

- "Hello, how are you?"
- "What can you help me with?"
- "Tell me about your capabilities"

Task-Specific Testing:

- Domain-specific queries for specialized agents
- Multi-turn conversations to test memory
- Complex reasoning tasks

Technical Testing:

- Long responses to test streaming
- Error scenarios to test robustness
- Attachment handling (where supported)

Last updated on

Overview

Understanding the Orbit Playground - a lightweight, standalone app for testing endpoints without authentication or admin features.

FAQ

Common questions and answers about using the Orbit App and Platform.