This project demonstrates how to instrument AI agents with Dynatrace to gain observability into Agentic AI workloads, including performance, cost, and runtime behavior.
By integrating Dynatrace with AI agents, developers can monitor agent execution, understand tool interactions, trace prompt and response flows, and analyze dependencies across distributed AI-driven systems.
Monitor AI agent interactions, tool usage, service dependencies, performance metrics, token consumption, and cost drivers—providing end-to-end visibility into how AI agent workflows behave at runtime.
Trace agent execution from the initial request through prompt flows, tool calls, and service interactions to the final response—enabling faster debugging and root cause analysis across complex agent workflows.
Use Dynatrace Intelligence to identify bottlenecks, optimize resource utilization, and understand how complex agent workflows behave across distributed services.
Add Dynatrace instrumentation to AI agents in minutes using simple integration patterns and practical examples—bringing runtime observability directly into agent development workflows.
This repository is designed for:
- Developers building AI-powered applications
- Platform and DevOps engineers operating AI systems
- AI practitioners working with agent frameworks
If you’re building AI agents, copilots, chatbots, or autonomous systems, these examples will help you add observability to your agent workflows and gain deeper insight into how your AI systems operate in production.
This repository includes examples and guidance for integrating Dynatrace with the following AI Agent building frameworks and libraries:
- Google Agent Development Kit (ADK)
- AWS Strands Agents
- OpenAI Agents SDK
- MCP - Model Context Protocol
- Anthropic - Claude Code
- GitHub Copilot SDK
If you’re using a framework that isn’t listed here, don’t worry! You can explore the Dynatrace Hub for the full list of supported technologies.