evlog

evlog is a TypeScript logging tool designed to provide structured, context-rich logs for a variety of environments, including frontend frameworks, API servers, background jobs, and browser clients. It addresses the challenges of observability by enabling developers to set and accumulate context, throw structured errors, and group recurring issues using typed catalogs. The tool supports capturing wide events that consolidate all relevant information, whether a request succeeds or fails, and offers actionable metadata such as root cause and fix suggestions alongside error logs.

A key aspect of evlog is its adaptability for both human and machine use. It features structured fields and machine-readable context, making it suitable for AI agents to analyze logs, diagnose issues, and suggest resolutions. The tool can write logs as NDJSON files locally, facilitating direct access for agents. Its pipeline architecture allows for batched log writes, automatic retries with exponential backoff, and fan-out to multiple destinations. Users can implement custom drain functions or use built-in adapters for services like Axiom, Sentry, PostHog, Better Stack, and the local file system.

evlog also supports client-side logging, capturing browser events and sending them to servers with automatic batching and retries. The system ensures delivery by switching to sendBeacon when the page is hidden and includes server-side validation, payload sanitization, and origin checks. Sampling is handled through a two-tier approach: head sampling filters logs by level, while tail sampling ensures critical events are retained based on customizable criteria.

For compliance and audit needs, evlog offers audit logging as an integrated layer over its wide events. Features include tamper-evident hash chains, redact-aware diffs, idempotency keys, and typed action catalogs. The schema covers typed actions, actors, targets, outcomes, and causation, supporting detailed and secure audit trails. The tool is positioned as a comprehensive logging solution for developers seeking structured observability and actionable insights across diverse environments.