Tim Pietrusky c4c8f9ccb2 feat(runpodctl): document --model-reference (model cache) + add eval scenarios (#21) ...

* test(runpodctl): add eval scenarios for skill behavior

Add 7 .eval.md scenarios covering autoscaling (v2.3 scale-by/scale-threshold),
pod auto-terminate, SSH connect, CPU pod creation, template+volume pods, Hub
serverless deploy, and image->template->serverless. The Hub scenario encodes
the GPU pool-id gotcha found via live testing (runpod/runpodctl#287).

* docs(runpodctl): document --model-reference (model cache) for serverless create

Adds the --model-reference flag (released in runpodctl v2.4.0) to the
serverless section: an example plus a 'Model cache' note covering the HF
cache path, the template-id + GPU-only constraints, and repeatability.
Verified end-to-end against the live API (create attaches the model and
resolves the ref to a commit SHA).

2026-06-19 15:10:25 +02:00

..

evals

feat(runpodctl): document --model-reference (model cache) + add eval scenarios (#21)

2026-06-19 15:10:25 +02:00

SKILL.md

feat(runpodctl): document --model-reference (model cache) + add eval scenarios (#21)

2026-06-19 15:10:25 +02:00