Changelog

What's shipping

A running, newest-first log of what changed in CleverQA.

  1. November 6, 2026

    Custom alerts: let your jobs report partial failures

    Custom alerts

    A heartbeat tells you a job didn't run. A custom alert tells you it ran and went wrong — for example a cron job that finished "successfully" but failed 8% of its records. Your own code reports the problem from its error branch. Read Your cron job said success. 8% of it failed..

    • A new Monitor service kind next to probes and heartbeats. Use both on the same job: the heartbeat catches silence, the custom alert catches trouble.
    • Trigger URL — POST a small JSON body to https://monitor.cleverqa.com/api/custom-alert/<token>, or to the same path without the token with an Authorization: Bearer <token> header. Every field is optional: key, severity (info, warning, error, critical), title and details (any JSON).
    • One alert per key — repeats count as occurrences on the same alert. Keys are free-form, so there is nothing to set up first.
    • Notifications — the first one arrives within about 30 seconds, then at most one summary every 10 minutes, only when there are new occurrences. Channels: email, SMS, Slack, Microsoft Teams and outgoing webhook destinations.
    • Details stay in CleverQA — notifications carry only the service name, a link and, in summaries, the count, never the payload. Obvious secrets (passwords, tokens, authorization headers, card-like numbers) are redacted before the payload is stored.
    • Every key files a defect in the linked project, and a closed defect reopens as a regression if its key fires again. Pushes to Jira, Linear, GitHub and Azure DevOps omit the payload.
    • Good to know — resolving an alert is manual, custom alerts don't appear on public status pages, and each one counts as one monitor toward your plan's limit. They're available on every plan, including Free.

    Monitoring

    • Richer heartbeat failure evidence — when a heartbeat goes quiet or reports a failure, the AI analysis now sees more of what was happening around it (failures that co-occurred, and a started-but-never-finished run), so the likely cause is less guesswork.
    • Load-test server alerts can file defects — when a stress run's resource alert fires (memory, CPU...), CleverQA can open a defect for it and close it when the alert clears. Opt in under Settings → Defect auto-create ("Server resource alerts", off by default).
  2. October 21, 2026

    Declare releases from CI, and weekday-only heartbeats

    Release declaration from CI

    • GitHub Action — add one step to a workflow and every release is declared to CleverQA, so crashes and incidents that follow attribute to it. CleverQA Release on the GitHub Marketplace.
    • Azure Pipelines task — the same for Azure DevOps: CleverQA Release on the Visual Studio Marketplace.
    • Both use an API token with the releases:write scope from Settings → API tokens, and re-running a pipeline updates the release instead of creating a duplicate. Declaring releases needs Essentials or above.

    Monitoring

    • Heartbeats on a weekly schedule — a heartbeat's expected interval can now be a time on chosen days of the week (for example Monday to Friday at 10:00 in your timezone) instead of a fixed duration. A job that only runs on weekdays is next expected on Monday after Friday's run, so weekends no longer raise false alarms.
  3. September 20, 2026

    Self-serve tokens, a public demo, free tools & security hardening

    Self-serve & workflow

    • API tokens — mint your own bearer tokens for CI and automation from Settings → API tokens. Scope a token to just what it needs, and revoke it anytime; the token value is only ever shown once, at creation.
    • Reconnect a crash integration — when a Crashlytics or Sentry connection's credentials expire, reconnect it in place from Settings instead of deleting and re-adding the integration.
    • Quality Score now weighs open defects — a release's Quality Score factors in its open defect count alongside monitor, crash, and stress signals, so it reflects outstanding work, not just live signal health.
    • Guided onboarding — new workspaces now get a four-step checklist — create a monitor, connect a crash source, declare a release, see your first diagnosed defect — that tracks real progress in your workspace instead of a generic product tour.

    New public pages

    Security & maintenance

    • Dependency and security hardening across the app, including upgrades to Next.js and NextAuth to pick up upstream CVE fixes.