What problem does it solve? Building production logging in Go with slog requires wiring together sampling, PII scrubbing, routing, HTTP middleware, and backend sinks — and mistakes like sampling after formatting, missing shutdown flushes, or silently dropped records are common. This Skill provides the canonical pipeline patterns and correct constructor usage for the 20+ samber/slog-* handler packages. ## Core Features & Use Cases - Pipeline Composition: Use slog-multi patterns (Fanout, Router, FirstMatch, Failover, Pool, Pipe) to route records to the right handlers with correct latency and error semantics. - Sampling & Formatting: Apply slog-sampling strategies (Uniform, Threshold, Absolute, Custom) as the outermost stage and slog-formatter middleware for PII masking and attribute transformation. - HTTP Middleware & Backend Sinks: Configure slog-gin/echo/fiber/chi middleware with per-status log levels, and connect sinks like Sentry, Loki, Datadog, Kafka, and Slack with proper graceful-shutdown flushing. - Use Case: Route ERROR logs to Sentry without sampling while sending sampled INFO logs to Loki, with PII scrubbed before any record leaves the process. ## Quick Start Set up a slog pipeline in Go using samber/slog libraries that samples INFO logs at 10%, scrubs PII, and routes errors to Sentry and everything else to Loki.