Open sourcePython · ≈19k stars · MIT
Agent Lightning
Microsoft's RL trainer that optimizes agents while they run inside their real harnesses.
Agent Lightning is Microsoft's framework for training AI agents with reinforcement learning. Version 1.0 (open-sourced August 2026) uses three components: a Trainer running 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. Agents keep their real harnesses, tools and control flow with no code changes. The project reports a Qwen3.5-9B coding-agent example improving SWE-bench Verified from 41.8% to 56.4% with 6K training samples.
In research (1)
- AgenticRoboRL: Bringing Coding-Agent Post-Training to Agentic RoboticsVerdict: Incremental
Oct 20, 2026 · Cited agent-training framework whose execution/optimization separation this builds on
Neighbours on the workbench
TypeScript · ≈393k stars · MIT
openclaw/openclaw
The viral open-source personal AI assistant that really does things.
Maintained by OpenClaw Foundation
TypeScript · ≈265k stars · MIT
deepseek-ai/deepseek-harness
Everything is a Plugin — DeepSeek's open agent harness.
Maintained by DeepSeek
TypeScript · ≈209k stars · Sustainable Use License (fair-code; GitHub reports NOASSERTION)
n8n-io/n8n
Fair-code workflow automation platform with native AI capabilities.
Maintained by n8n-io
Connections
Relations from the Harness Hub knowledge graph, as charted in the State of Harness reports.
Maintained by · 1
- MicrosoftOrganisation
Similar entries · 1
- OpenPipe (ART)Startup
Sources
Verified Oct 29, 2026