Raycaster/ Eval

Domain evidence · Model development and data

Environments and evaluation records for post-training.

Browse the full public catalog; private holdouts stay gated.

Flagship traces · including failures

Evidence depth first. Not a leaderboard.

Includes runs that miss rubric criteria—failure is part of the product.

Workflows

Stay inside the work that brought you here.

Environment and data program

Representative tasks, graders, and QC.

Measured failure modes

What current agents miss in this domain.

  1. Leaderboard scores without trajectory evidence

Why Eval exists · why Workspace exists

Public evidence and cloud agents are the same harness.

Eval exists so scores are inspectable—tasks, trajectories, artifacts, and rubric verdicts anyone can open.Workspace exists so people can automate real file work with that harness, and so Raycaster never evaluates work it cannot perform.

01Stay in the evidenceOpen more traces, domains, and verdicts on this site—the public proof layer.Browse domains →02Use the cloud workspaceWhen you want to run agents on files yourself, Workspace uses the same harness.Start in Workspace →