Zum Inhalt springen
Kitbay

launchdarkly-online-evals

Apache-2.0Aktualisiert am 11. Nov. 2026138GitHub-Sterne

A partner-maintained skill for AWS DevOps Agent, from LaunchDarkly, for automatic quality scoring of AI config output in LaunchDarkly using judges, where one model rates another.

Was der Skill vermittelt

A partner-maintained skill for AWS DevOps Agent, from LaunchDarkly, for automatic quality scoring of AI config output in LaunchDarkly using judges, where one model rates another. It covers the built-in judges, creating custom judges, attaching them to config variations, setting sampling rates and watching quality scores, and notes that judges work only in completion mode. It needs the hosted LaunchDarkly MCP server, and SDK examples use the Python and Node.js AI SDKs.

Funktionen

  • Attach judges to AI config variations
  • Create custom judges
  • Set evaluation sampling rates
  • Monitor quality scores