The Forge
What it is
The harness is what runs the studio's agents. A run moves through a governed lifecycle: the plan is checked by an independent step before any work starts, the finished work has to clear two separate verification bars before it lands, and everything in between writes to an audit spine that can be read back afterwards. None of that makes an agent clever. It makes an agent's failures visible early, which in practice matters more.
The open-source intent
The intent is to open-source it, as a harness evaluated properly for near-frontier models and small local models. There's no date attached, and there won't be one until the evaluation numbers justify it.
What "heavily evaluated" means
The same treatment the harnesses in the study get: the same task sets, deterministic grading, published intervals. When this harness has results of that kind, they'll be on the research pages, whether or not they flatter it.