Skip to content

Changelog

Notable, user-facing changes. Dates are the release/merge date. This is a curated summary — the per-PR history lives in the repo's commit log and pull requests.

Unreleased — v1.1 (in progress)

Portability, auto-monitors, and polish on top of v1:

  • dbt as a third orchestration provider — observe dbt builds and trigger suites on success, via a post-build HMAC callback + a run_results.json artifact poll (ADR 0029).
  • Apache Iceberg as a fifth datasource — native pyiceberg read (v2 baseline; ADR 0030, #716/#722), with natively-computed freshness/volume monitors, a column profiler + column listing, and dry-run preview (#721).
  • Freshness & volume monitors — the first auto-monitor kinds (is the data stale? did the load land whole?) on SQL datasources plus Iceberg.
  • Vendor-neutral observability — OpenTelemetry logs + traces, exportable to Application Insights and/or a generic OTLP endpoint.
  • Personal access tokens (PATs) — dq_live_ tokens for headless / AI-client use, on REST and MCP (ADR 0026).
  • Workspace-admin visibility extended to the MCP tools + schedules; dry-run preview extended to every datasource.
  • Run failure reasons — a run that fails to execute now shows a redaction-safe reason.
  • Secret lifecycle — connection delete cleans up its stored secret.
  • Assets as the primary lens — a data asset (table/file) is now a first-class entity with its own page: health rolled up across every suite that targets it, open incidents, and lineage. The dashboard and sidebar lead with assets; suites and runs link back to the asset they touch (ADR 0034).
  • Lineage graph — an asset's provenance and blast radius render as one left-to-right graph, one column per hop, with clickable nodes.
  • Asset browse by source — drill down datasource → database → schema → table, with a flat searchable table as the second lens.
  • Two health axes on an asset — "could DataQ reach the datasource?" is shown separately from "is the data good?", so an unreachable datasource no longer masquerades as a data failure, and a run that evaluated nothing no longer reads as a green pass.
  • Datasources read as names — the UI shows Snowflake · ACCT / ADLS · account/container / iceberg_catalog instead of the raw connection string (an Iceberg namespace is a full DSN). The raw identifier stays available on hover and via copy.
  • Mobile — the sidebar becomes an overlay drawer on a narrow viewport, and the share / edit panels reflow so their controls stay on screen (previously the "Add" button in the share drawer was painted off the right edge, making a suite unshareable from a phone).
  • A broken orchestration poll now tells you — an expired credential silently stopped pipeline-run ingest, suite triggering, and lineage refresh for six days in our own production. A failing poll is now a fact about the connection: the connections list badges it with a failure count, the lineage panel warns instead of showing a confident empty graph, and after 3 consecutive failures an alert is pushed to the workspace channel — once on the way down and once on recovery, never once per poll. The reason is always the classified one, never raw error text (the real failure carried a SAS token in its message).

Fixed

  • Iceberg catalog credential exposure — the SQL-catalog password had to be carried inline in the catalog_uri (the connection type had only one secret slot), which meant it was persisted in the connections table, copied into the asset identity, returned by the API, and rendered in the UI. Iceberg connections now take a second secret (catalog_secret_name), the config rejects a password in catalog_uri, and a migration scrubs existing rows. Action required: an existing Iceberg connection whose catalog_uri carried a password must be re-pointed at a secret and the password rotated — treat it as disclosed.

  • /assets deep links 404'd in the production image — the nginx rule for the hashed bundle directory also swallowed the /assets app route.

  • Docs: Features overview, Recommended usage, this changelog, security, REST API, troubleshooting, deployment, and a first-suite tutorial.

v1.0.0 — 2026-07-04

First production release. Deployed to Azure Container Apps.

  • Data-quality checks across four datasources — Snowflake, Unity Catalog, ADLS Gen2, AWS S3 — on Great Expectations, with a catalog-driven check editor and custom SQL.
  • Suites with per-suite view/edit sharing and export/import for env promotion.
  • Runs — run-now with live progress + cancel, cron scheduling, and pipeline triggers from ADF & Airflow.
  • Results & dashboard — severity tiers, a health score, trends, and column-aware redacted failing-row samples.
  • Alerting — Teams / Slack / email with severity routing, dedup, and per-check snooze.
  • AI assistants — an 8-tool MCP server for Claude / Copilot / Cursor.
  • SSO (OIDC) and secrets in a managed vault; the frontend is the sole public surface.

See the Features page for the full current capability set.