AgentOps

Tools & Infrastructure 💻 Python ⚖️ MIT 🟡 Quiet lately
5.8k stars

AgentOps is an open-source observability and development platform for AI agents, providing session replays, analytics, and debugging tools from prototype to production. It helps developers build, evaluate, and monitor AI agents by automatically tracking LLM calls and offering integrations with popular frameworks. It is designed for developers working with AI agents who need visibility into agent behavior and performance.

AgentOps demo
🖼️ Screenshot from the project README

✨ Key features

  • Replay analytics and debugging with step-by-step execution graphs
  • LLM cost management tracking spend across providers
  • Native integrations with CrewAI, AG2, LangGraph, and more
  • Self-hosting option for running the full app on your cloud
  • Decorators for sessions, agents, operations, and workflows
  • Supports async, generators, and custom attributes

🎯 Use cases

  • Monitor and debug multi-agent systems built with OpenAI Agents SDK
  • Track costs and performance of CrewAI agent crews
  • Add observability to AG2 (AutoGen) agents with two lines of code
  • Analyze CAMEL agent interactions and tool usage
  • Instrument Langchain applications with callback handlers

📦 Installation

🧰 Requirements: Python 3.x or Node.js for TypeScript, API key from AgentOps dashboard, and optional API keys for LLM providers (e.g., OpenAI, Anthropic).

pip install agentops

For specific integrations, install additional packages as needed, e.g., pip install 'crewai[agentops]' or pip install agentops[langchain].

🚀 Usage

import agentops

# Beginning of your program (i.e. main.py, __init__.py)
agentops.init( < INSERT YOUR API KEY HERE >)

...

# End of program
agentops.end_session('Success')

❓ FAQ

How do I get an API key?

You can get an API key from the AgentOps dashboard at https://app.agentops.ai/settings/projects.

Can I self-host AgentOps?

Yes, the full AgentOps app (Dashboard + API backend) can be run on your own machine; see the setup guide in app/README.md.

Which frameworks are supported?

AgentOps has native integrations with CrewAI, AG2 (AutoGen), Agno, LangGraph, OpenAI Agents SDK, Langchain, Cohere, Anthropic, Mistral, and Camel AI.

How do I track a specific operation?

Use the @operation or @task decorators from agentops.sdk.decorators to create spans for tracking specific operations.

📊 Repository

Stars★ 5,809
Forks🍴 617
Open issues🐛 177
Last commit🕒 Jun 25, 2026
Created📅 Aug 2023
Language💻 Python
License⚖️ MIT

🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.