Handles the full lifecycle of custom business metrics in LaunchDarkly's AI Config system. You define metrics via their API (revenue, satisfaction scores, task success rates), track events through the SDK with metric_value parameters, then query the results. The skill includes auto-detection of API keys from Claude's MCP config or environment variables before prompting, which is a nice touch. The examples are genuinely useful: tracking conversions with amounts, negative feedback alerts when satisfaction drops below 50, revenue attribution after AI interactions. Requires a LaunchDarkly API token with writer role. This is for teams already using LaunchDarkly who need custom metrics beyond the built-in AI observability stuff.
npx skills add https://github.com/launchdarkly/agent-skills --skill aiconfig-custom-metrics