Jitsu Changelog

Everything you can see in Live Events — function logs, warehouse batch statuses, streaming delivery results, and dead-lettered events — can now be exported to your own monitoring stack in near real time. Records are delivered as OpenTelemetry (OTLP) log records over HTTP, so they work with Datadog, Grafana Cloud, New Relic, Elastic, or any OpenTelemetry Collector.

• Alert on delivery errors and failing functions with the tools your team already uses.
• Keep operational logs beyond the Live Events retention window.
• Filter by workspace, connection, destination, or record type via per-record attributes.

Available on Enterprise plans of Jitsu Cloud.

Our first stable public release in nearly a year, and the biggest self-hosting upgrade since Jitsu Next. Functions, Profile Builder v2, and connector syncs now run as managed Kubernetes workloads, reconciled by a new operator service — with autoscaling, workload classes, and no Google Cloud Scheduler dependency.

• Audit log (SOC2-oriented) with account-activity alerts, plus named user API tokens with expiration.
• Dead-letter queue with a reprocessing worker: failed events can be replayed instead of lost.
• Redesigned signup with server-side email verification and improved OIDC support.
• Development Helm chart: deploy the full architecture to Minikube with zero configuration.
• Shipped alongside: jitsu-cli 1.11 with a config command tree, OpenAPI spec dump, and default workspace support.

Breaking: a feature-complete production deployment now requires Kubernetes; docker-compose remains for exploration and development. Read the self-hosting docs before upgrading.

Jitsu 2.14 also grew the integrations catalog:

• New destinations: Resend and SendGrid (email), Statsig (feature management), and DuckDB / MotherDuck (warehouse).
• New Jitsu-maintained Xero source connector.
• Snowflake key-pair authentication, and much faster Snowflake merges on large tables.
• Redshift IAM role-based authentication and SUPER-type support for JSON columns.
• Postgres destination via Google Cloud Private Service Connect.
• Firebase source: subcollection sync via CollectionGroup queries.

The Jitsu console now exposes a Model Context Protocol server with OAuth 2.1 and Dynamic Client Registration. Point Claude, Cursor, or VS Code at it and your agent can configure destinations and streams, watch Live Events in real time, and edit the Functions that transform your data — build and debug, end to end.

• GTM destination: a new loadGtm option lets you load Google Tag Manager yourself — Jitsu then only pushes to the data layer — plus a resetDataLayer option that clears only Jitsu-set keys between events.
• preInitAnonymousId creates the anonymous-id cookie synchronously at init, so it's available for first-page server-side reads.
• Passing userId in the options now auto-identifies on startup, and the new user() method returns the current, privacy-aware user.
• Fixed: array values in payloads now replace instead of merging element-by-element, server-side events without a page URL no longer throw, anonymous-id cookies scope correctly on vercel.app / webflow.io, and GA4 skips malformed page_view events.

Your data pipeline shouldn't live behind a UI. The console API — including the Sync API — is now described by a public OpenAPI spec with a built-in interactive reference, and the new jitsu-cli manages streams, destinations, connections, and functions from scripts and CI. The same API surface is what AI agents drive end-to-end.