logging-discipline

What problem does it solve?

It prevents incident-blind, unsearchable log firehoses and reduces the risk of leaking secrets or PII by enforcing structured logging, level discipline, correlation IDs, PII redaction, and log volume budgeting.

Core Features & Use Cases

• Structured-over-free-text logging with event identifiers and key-value fields for machine filtering and aggregation.
• Level discipline (ERROR/WARN/INFO/DEBUG/TRACE) to keep production signal usable and avoid turning INFO into noise.
• Correlation IDs (request ID, trace ID, user/session hashes) so incidents can be reconstructed across services and async boundaries.
• PII redaction at the logger layer using configurable redaction/hashing rules rather than scattered call-site fixes.
• Log volume budgeting and sampling using request-consistent sampling (and optional tail-based sampling) to control cost without losing debuggability.
• 2am diagnosis filter to ensure each log line has an event, context, reason, and actionable content.

Quick Start

Use the logging-discipline skill to design your service’s logging scheme so every line is structured, correlatable, redacted, and volume-budgeted.