Welcome to AuraOne Open

[gchahal1982]

This is the public side of AuraOne — open tools for the human-judgment layer of AI evaluation.

The bet behind this release: the best evals will stay private. The standards for building trustworthy evals should not.

What lives here today:

• EvalKit — a Python package and CLI for rubric validation, scoring, judge calibration, reviewer agreement, drift, and leakage audits. Runs locally. No account, no tenant, no network.
• Robotics ReviewKit — schemas, exporters, taxonomy, and a static viewer for teleop review and failure data.
• Buying Toolkit — templates and checklists for teams buying human-data work.

What lives here is intentionally narrow. Tutorial datasets are synthetic and disclosed as such. Robotics examples are mock metadata. There is no expert-authored benchmark in this release. The release ships the standards, not the dataset.

Use Discussions for questions, rubric examples worth sharing, exporter coverage requests, and ideas for the next release. Use Issues for bugs and concrete proposals.

— AuraOne