AI prompts repo
Ship prompts like production software.
Create prompt versions, select real models from the chosen provider, test output behavior, enforce eval gates, and promote approved versions across environments through AI change management.
Dynamic model selectionThe default model follows the selected provider and comes from the live provider model catalog, with a dropdown for testing prompts across supported models.
Registry workflow
CreateDefine reusable prompts with variables, owner metadata, and provider-aware default models.
VersionManage draft and published versions so released prompts stay immutable and traceable.
TestRender prompts and test against selected provider models before release.
EvaluateRun eval suites and block publishing when quality gates fail.
DeployPromote versions to development, staging, and production environments.
ObserveConnect prompt usage, render logs, request traces, and release outcomes.