Flight Tracker Agent (OpenAI SDK + uAgents)

FlightStatusAgent (Flight Tracker)

Fetches live flight status (Flightradar24 / FlightAware / Flight.info) using OpenAI’s Agent SDK, then connects to uAgents and Agentverse so your flight tracker can be discovered and chatted with by ASI/ASI:One.

Check out the full example here: FlightStatusAgent (Flight Tracker).

Overview

This repo demonstrates how to:

• Build an OpenAI Agent via Agent Builder
• Use the Agent SDK (Python) to define and run it programmatically
• Bridge it into the uAgent Protocol (so it works on ASI Agentverse)
• Publish the agent so that it’s discoverable by ASI/ASI:One

Architecture

User → ASI / Agentverse → uAgent Bridge → OpenAI FlightStatusAgent → Web Search Tool (Flightradar24 / Flight.info)

Core components:

• workflow.py — defines the OpenAI agent logic
• uagent_bridge.py — connects the agent to Agentverse using uagents
• .env — contains your OpenAI API key
• requirements.txt — pinned dependencies

1. Build the FlightStatusAgent on OpenAI

• Go to Agent Builder
• Create a new agent, name it Flight Tracker, and paste these instructions:

You are FlightStatusAgent. Your job is to fetch today’s live flight status for the flight number mentioned by the user, using Web Search to find up-to-date data (Flightradar24, FlightAware, Flight.info, etc.).

Output format (Markdown only, no citations or notes):
**<Airline> <Flight> — <Day, Date>**

- **Status:** <status>  
- **Route:** <Origin (IATA)> → <Destination (IATA)>  
- **Local Times:**  
  - **Departure (scheduled):** Oct 13 2025 20:55 EDT (JFK)  
  - **Departure (actual):** Oct 13 2025 23:16 EDT (JFK)  
  - **Arrival (scheduled):** Oct 15 2025 07:45 NZDT (AKL)  
  - **Arrival (actual):** <if available>  
- **Aircraft:** <model>; registration <reg>  
- **Terminals / Gates / Baggage:** <info if available>  
- **Sources:**  
  - [Flightradar24](https://www.flightradar24.com/data/flights/{flight_num})  
  - [Flight.info](https://www.flight.info/{flight_num}?utm_source=openai)

Rules:
- Never show citations or reference markers.  
- Use 3-letter timezone abbreviations (PST, EDT, NZDT).  
- Omit missing fields instead of guessing.  
- Always include Flightradar24 link.  
- Keep the output concise and visually clean.

• Add the Web Search Tool under Tools.
• Set model to gpt-5 and reasoning effort to medium.
• Save and deploy — the Agent Builder will give you a workflow ID.

2. Use the OpenAI Agent SDK

• Your workflow.py (see sample in this repo) defines:
• The agent’s code (Agent, WebSearchTool, ModelSettings) -The async runner run_workflow() that wraps your agent execution

Run it locally:

python3 workflow.py

3. Bridge the Agent with uAgents

The file uagent_bridge.py registers your OpenAI agent as a uAgent:

• Responds to messages using the AgentChatProtocol
• Passes user flight queries to your OpenAI workflow
• Returns the markdown output to ASI clients

Run it locally

python3 uagent_bridge.py

You’ll see logs such as:

INFO: [FlightTracker]: Agent inspector available at https://agentverse.ai/inspect?...
INFO: [FlightTracker]: Mailbox client started

Use inspector link to connect agent to agentverse using mailbox. That means it’s successfully published on the Agentverse network.

4. Add .env and requirements.txt

.env

OPENAI_API_KEY=your_openai_api_key_here

requirements.txt

openai
uagents
pydantic
python-dotenv

5. Example Interaction

On ASI.One or any connected LLM:

@flight-tracker NZ1 status