online-evals
launchdarkly1.8k installs
Quality scores for your LaunchDarkly config variations.
What it does
Configures response-evaluation judges and sampling rates so users can monitor quality scores for LaunchDarkly config variations.
When to use it
Use when attaching judges to config variations for automatic response-quality scoring.
How to use it
Give it a judge-creation or attachment request; it creates or updates judge configurations in LaunchDarkly.
Uses
Good to know
- cannot be undone: LaunchDarkly; your model provider
- sends a credential to a server: API access token
Latest review
What you need · 4
Requires a LaunchDarkly account with AgentControl enabled.
Enables the preferred OAuth-authenticated management route without an API access token; direct REST calls are the fallback.
Access · 2
This skill
LaunchDarkly
Write
Updates LaunchDarkly judges and targeting (irreversible)
your model provider
Write
Sends paid model evaluation requests (irreversible)