Verified project record

microsoft/agent-lightning

Agent Lightning is a Python library that lets you train and optimize AI agents using reinforcement learning, automatic prompt optimization, and supervised fine-tuning with almost zero code changes. It works with existing agent frameworks including LangChain, AutoGen, CrewAI, and the OpenAI Agent SDK.

Project overview

The library addresses the problem of agent optimization typically requiring extensive code changes and framework lock-in by supporting multiple training algorithms across any agent framework.

Project type: AI Agent · Infrastructure
Use cases: Automation
Deployment: Refer to project documentation
License: MIT

Best for

AI engineers, developers, and researchers who need to apply training algorithms to existing Python-based AI agents without extensive code rewrites or framework migration.

Key capabilities

Turn your agent into an optimizable target with almost zero code changes.
Build with any agent framework or even without an agent framework.
Selectively optimize one or more agents in a multi-agent system.
Supports algorithms including reinforcement learning, automatic prompt optimization, supervised fine-tuning, and more.

Limitations and risks

The library is Python-based and deployed via pip.
GPU requirements, data boundaries, cost dependencies, and telemetry are not documented.

Getting started

Installation requires a single pip command: pip install agentlightning. Coding is required to use the library.

Evidence and sources

GitHub project description: The absolute trainer to light up AI agents.
README: Turn your agent into an optimizable beast with **ZERO CODE CHANGE** (almost)! 💤
README: Build with **ANY** agent framework (LangChain, OpenAI Agent SDK, AutoGen, CrewAI, Microsoft Agent Framework...); or even WITHOUT agent framework (Python OpenAI). You name it! 🤖
README: Embraces **Algorithms** like Reinforcement Learning, Automatic Prompt Optimization, Supervised Fine-tuning and more. 🤗
README: pip install agentlightning

Source updated at: 2026-07-16T18:03:58Z

Open original GitHub project