🪢 Open source agent evals & observability: Trace, evaluate, and improve LLM applications with one open platform.
TypeScript analyticsautogenevaluationlangchainlarge-language-modelsllama-indexllmllm-evaluationllm-observabilityllmopsmonitoringobservabilityopen-sourceopenaiplaygroundprompt-engineeringprompt-managementself-hostedycombinator
/**
* Pure decision shared by the client capability hook (`useEvalCapabilities`)
* and the server create-gate (`assertCanCreateLegacyEvalJob`): may a *new*
* evaluator use the legacy (trace/dataset) experience?
*
* - events_only: legacy tables are no longer written → no new legacy evals.
* - dual: self-hosted deployments always allow legacy; on Cloud, no new legacy.
* - legacy: legacy is the only experience.
* - forced-v3 projects keep the legacy experience unless events_only is set.
* - undefined write mode (session still loading client-side) → not allowed,
* which keeps legacy options hidden until the mode is known.
*/
export function isNewLegacyEvalAllowed(params: {
v4WriteMode: "legacy" | "dual" | "events_only" | undefined;
isLangfuseCloud: boolean;
isForceV3Project: boolean;
}): boolean {
const { v4WriteMode, isLangfuseCloud, isForceV3Project } = params;
if (v4WriteMode === "events_only") return false;
if (isForceV3Project) return true;
if (v4WriteMode === "dual") return !isLangfuseCloud;
return v4WriteMode === "legacy";
}