auraone-evalkit v0.2.1

SEO/readme/package metadata refresh for AuraOne EvalKit. The package README now leads with the local CLI outcome, install command, copy-paste quickstart, concrete use cases, comparison notes, trust signals, and clearer compatibility limits. PyPI metadata now includes the optimized summary, keywords, changelog, security, and CI links.

v0.1.1 — Initial public release (packaging fix)

AuraOne Open v0.1.1

Initial public release of AuraOne's open tools for the human-judgment layer of AI evaluation.

Note on 0.1.0: v0.1.0 was published to PyPI without the auraone_evalkit.reports subpackage due to a .gitignore pattern that shadowed the source directory. 0.1.1 is the first installable release. Use pip install auraone-evalkit==0.1.1.

What's included

• EvalKit (packages/evalkit/) — auraone-evalkit Python package + evalkit CLI for rubric validation, linting, scoring, judge calibration, reviewer agreement, drift detection, leakage audits, sampling, weight calibration, reporting, and dataset cards. 58 tests.
• Robotics ReviewKit (robotics-reviewkit/) — teleop review schema, intervention ontology, failure-mode taxonomy, LeRobot + RLDS/OpenX exporters, static viewer, mock example episode. 26 tests.
• Buying Toolkit (resources/buying-toolkit/) — SOW, SLA, RFP, pilot design, vendor comparison, reviewer certification, and program playbook templates.
• Writing (resources/writing/) — the thesis posts behind this release.

Install

```bash
pip install auraone-evalkit
```

Hosted artifacts

• PyPI: https://pypi.org/project/auraone-evalkit/0.1.1/
• Hugging Face datasets:
  • https://huggingface.co/datasets/auraoneai/evalkit-tutorial-v0.1
  • https://huggingface.co/datasets/auraoneai/synthetic-multiturn-eval-v0.1
  • https://huggingface.co/datasets/auraoneai/robotics-reviewkit-mock-episodes-v0.1
• Website: https://auraone.ai/open and https://auraone.ai/resources

What this release is not

• Not an expert-authored benchmark. Tutorial datasets are synthetic and exist to make the tooling runnable.
• Not real robotics data. The example episodes are mock metadata.
• Not a hosted AuraOne service. EvalKit runs locally with no API key.

License

MIT.