FULL STORY
Harness²: Google's Recursive Self-Evolving Agent Goes Open Source
Google, with Northwestern and Google Cloud AI Research, unveiled Harness², a recursive self-evolving agent framework that improves agents without model training. Days later, the project was open-sourced under Apache-2.0 and quickly gained traction on GitHub.
2026-10-14 ~ 2026-10-18 · 2 episodes · 6 posts
Episode 1 · Google's Harness²: Recursive Self-Evolving Agent Framework (2026-10-14, 2 posts)
Google, with Northwestern and Google Cloud AI Research, published Harness², a recursive self-evolving agent harnessing framework that improves performance by up to 11.8 points without requiring a verifier.
Episode 2 · Google Open-Sources Harness² for Recursive Agent Self-Optimization (2026-10-18, 4 posts)
Google Research open-sourced Harness², an Apache-2.0 framework that lets agents recursively optimize their own execution harness without retraining or modifying model weights, delivering gains of up to 11.8 points.
- Google open-sources Harness²: recursive agent harnessing lifts GLM-5 and Gemini 3.5 Flash by up to 11.8 points — dr_cintas · 2026-10-18
- Google Research open-sources Harness², a recursive agent harnessing framework (up to +11.5 pts) — Saboo_Shubham_ · 2026-10-18
- Google Open-Sources Harness², an Agent Framework That Improves Its Own Harness — algo_diver · 2026-10-18
- Google open-sources Harness², letting agents recursively optimize their own harness without touching model weights — aigclink · 2026-10-18