feat(integrations): Implement OpenAI Agents SDK Integration

[viniciusdsmello]

Implement OpenAI Agents SDK Integration with Example Notebook

Description

This PR implements a complete OpenAI Agents SDK integration for the Openlayer Python SDK and provides a comprehensive Jupyter notebook example demonstrating its usage. This enables automatic tracing and monitoring of multi-agent workflows built with the OpenAI Agents SDK.

What's Implemented

🔧 New Integration: src/openlayer/lib/integrations/openai_agents.py

Complete OpenAI Agents SDK Integration:

• ✅ OpenlayerTracerProcessor - Main tracing processor class
• ✅ Automatic span capture - Traces all agent interactions, handoffs, and tool calls
• ✅ Conversation flow tracking - Maintains context across agent transitions
• ✅ Performance monitoring - Captures latency, token usage, and metrics
• ✅ Error handling - Comprehensive error capture and reporting

Key Features:

• Agent workflow tracing - Complete visibility into multi-agent conversations
• Tool call monitoring - Tracks function tool usage and outputs
• Handoff tracking - Captures agent-to-agent transitions
• Context preservation - Maintains conversation state across agents
• Token usage tracking - Monitors LLM token consumption
• Response capture - Extracts actual LLM outputs and agent responses

📓 Example Notebook: examples/tracing/openai/openai_agents_tracing.ipynb

Comprehensive Tutorial & Demo:

• ✅ Google Colab compatible - Ready to run in Colab with installation instructions
• ✅ Step-by-step tutorial - From setup to advanced multi-agent workflows
• ✅ Real-world example - Airline customer service with 3 specialized agents
• ✅ Production patterns - Best practices for agent architecture
• ✅ Interactive examples - 4 different test scenarios

Demonstrates Integration Usage:

1. Environment setup - API keys and dependencies
2. Tracing configuration - Setting up OpenlayerTracerProcessor
3. Agent creation - Triage, FAQ, and Seat Booking agents
4. Function tools - Reusable tools for different tasks
5. Agent handoffs - Seamless transfers between specialists
6. Hook functions - Custom logic during handoffs
7. Conversation flows - Complete customer interactions
8. Monitoring results - What appears in Openlayer dashboard

Testing & Validation

• Integration captures all agent interactions correctly
• Notebook executes successfully in Colab and locally
• Agent handoffs are traced properly
• Tool calls are monitored accurately
• Performance metrics are captured
• Error handling works as expected
• Openlayer dashboard displays traces correctly

The integration makes it trivial for developers to add world-class monitoring to their OpenAI Agents SDK applications, improving both development velocity and production reliability.