Skip to content

v25.1.0

Choose a tag to compare

@LaunchDarklyReleaseBot LaunchDarklyReleaseBot released this 27 Oct 16:06
Immutable release. Only release title and notes can be modified.
5d51188

Added

  • Added optional judgeScores to AgentControl config metrics. It lists the average score and sample count for every judge metric recorded in the time range, keyed by the judge's evaluation metric key ($ld:ai:judge:<suffix>), including custom judges and the built-in accuracy, relevance, and toxicity judges. It is omitted when no judge scores were recorded. The existing judgeAccuracy, judgeRelevance, and judgeToxicity fields are unchanged. Affected endpoints:
    • GET /api/v2/projects/{projectKey}/ai-configs/{configKey}/metrics
    • GET /api/v2/projects/{projectKey}/ai-configs/{configKey}/metrics-by-variation
  • Added optional analysisUnit on each metric in environment release responses. It is the context kind used as the metric's analysis unit. It is the release's randomization unit unless the metric uses a decoupled unit of analysis. Releases started before analysis units were recorded omit analysisUnit, and those metrics are analyzed on the randomization unit. Affected endpoints:
    • GET /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases
    • POST /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases
    • GET /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases/{releaseId}
    • PUT /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases/{releaseId}/stop
  • Added optional analysisUnit on each metric in POST /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases. It sets the context kind used as the metric's analysis unit. Omit it to analyze on the randomization unit. A value must differ from the randomization unit and be one of the metric's units. Percentile metrics, ratio metrics, and metrics in a group cannot set it.