microsoft/AIOpsLab

What is it?

Key features

• 01Deploys microservice applications to a Kubernetes cluster via Helm or kubectl.
• 02Injects faults and generates workloads using built-in injectors and wrk2-based load generation.
• 03Provides a benchmark suite of detection, localization, analysis and mitigation problems.
• 04Lets agents plug in through a simple async get_action interface and a shared orchestrator.
• 05Records agent sessions and evaluation results for later analysis.

Use cases

• Benchmarking LLM-based agents on incident detection and mitigation in a realistic microservice setup.
• Developing and testing an AIOps agent before running it against production-like clusters.
• Adding custom applications, faults or evaluation metrics to build a reproducible internal benchmark.

Quick start

1. 1

   Install Python 3.11

   sudo apt install python3.11 python3.11-venv python3.11-dev -y
2. 2

   Install Poetry with the official installer from the README

3. 3

   Clone the repository with submodules (replace the placeholder with the repo URL)

   git clone --recurse-submodules <CLONE_PATH_TO_THE_REPO>
4. 4

   Install the project in editable mode with pip from the repository root

   pip install -e .

Things to know

• !Requires a Kubernetes cluster, such as a local kind cluster or a remote cluster, plus Helm.
• !Some fault injectors need Docker on the local machine, and full functionality requires running on a controller VM.

In collections

Topics

FAQ

What is AIOpsLab?

A Python framework for designing, developing and evaluating autonomous AIOps agents on simulated microservice cloud environments with fault injection. AIOpsLab is a framework for building and benchmarking agents that handle IT operations tasks such as detecting, localizing, analyzing and mitigating faults in cloud services. It can deploy microservice applications on a Kubernetes cluster, inject faults, generate workloads and export telemetry data, while orchestrating these components and exposing interfaces for agents to interact with the system.

Is AIOpsLab free to use? What license does it have?

AIOpsLab is open source under the MIT license. Check the license terms before using it in commercial products.

Is AIOpsLab still maintained?

Actively maintained. The last commit to AIOpsLab was on Nov 6, 2026.

What can I use AIOpsLab for?

Benchmarking LLM-based agents on incident detection and mitigation in a realistic microservice setup. Developing and testing an AIOps agent before running it against production-like clusters. Adding custom applications, faults or evaluation metrics to build a reproducible internal benchmark.

How do I install AIOpsLab?

Install Python 3.11: sudo apt install python3.11 python3.11-venv python3.11-dev -y Install Poetry with the official installer from the README Clone the repository with submodules (replace the placeholder with the repo URL): git clone --recurse-submodules <CLONE_PATH_TO_THE_REPO> Install the project in editable mode with pip from the repository root: pip install -e .

Where can I get AIOpsLab?

The source code and installation instructions are on GitHub: https://github.com/microsoft/AIOpsLab