Know whether your tuned checkpoint is ready to ship.
Evaluates a trained checkpoint through promotion gates and produces an evidence-backed PROMOTE or REJECT report.
When to use it
Use after training produces a checkpoint, when deciding whether a tuned model ships, or when re-gating against updated goldens.
Give it a trained checkpoint, baseline results, and frozen drift suite; it produces a promotion report committed alongside the checkpoint.
What you provide
No setup requirements listed.
This skill
production rollout
Starts a production canary rollout