ATLAS is open source under AGPL-3.0. It sits between the agent and inference server, allocates compute by task difficulty, checks every attempt, and returns what held up. It needs 16 GB of VRAM, around 20 GB of free disk, and Docker or Podman already installed.
The scorer reads the serving model's hidden states and sets the candidate count and reasoning budget. It rejects failures, unresolved cross-file calls, and code that runs without doing the task. If no candidate survives, the runtime repairs against the recorded failures and runs the suite again.
It provides per-stage record of what was consumed.
No comments yet.