Skip to content
Agentmart
Submit a tool

agent-lightning

Reinforcement-learning trainer of about 3,500 lines that trains models behind agents run in their own harnesses, locally or as Kubernetes jobs.

Harness by microsoft

IndexedMIT★ 18,800
Write a review

Chalk Labs

Need help deploying agent-lightning?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Get started

Setup steps are in the repository. Website · Repository

About

Agent Lightning v1.0 is a lightweight framework for training agents with reinforcement learning while they keep running in their real harnesses. It has three components: a trainer that runs verl and vLLM, an API gateway that proxies model requests and captures training data, and a rollout controller that runs agents locally or as Kubernetes Jobs. The README reports coding-agent examples that raise SWE-bench Verified scores on Qwen3.5 models. It is a training tool rather than an agent-building framework.

Use cases: model-inference

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Fact sheet

Languages
PythonREADME
Deploys to
kubernetesREADME
Maturity
Stablegithub.com

Facts extracted by Claude Sonnet 5.5 from the linked sources on Oct 24, 2026; reviewed Oct 24, 2026. Wrong fact? Tell us.

Alternatives

Frameworks for the same language with a similar way of orchestrating agents.

Signals

Stars
18,800
Forks
1,693
Open issues
104
Last commit
Oct 20, 2026 (11 days ago)
Last release
Sep 29, 2026 (1 month ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from microsoft

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get agent-lightning running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying agent-lightning?

CallGet help
Cite this page

agent-lightning. Agent Mart. https://agentmart.dev/harnesses/agent-lightning (last updated 2026-10-24)