OPEN-SOURCE PROJECT

tenbin

Documentation, an MCP server and a Skill for designing Jev judgments with question linting, batch evaluation and calibration.

Stars
2
Forks
0
License
MIT
Last commit
2026-09-21

What Jev does here

Runs Choice, Score and Noul over samples and uses measured results to design local thresholds.

Connects question design, sample evaluation and runtime rules.

Calibration requires labeled samples for the workload and does not make judgments infallible; tests were not rerun here.