Alert inbox: mark as read, delete, and plan retention

Alert inbox: mark as read, delete, and plan retention

Slack is a terrible archive. Threads bury the failed invoice, guests cannot search last month, and nobody agrees what “done” means. AlertHub stores every ingested alert in the project feed first — then optionally pages Slack or Telegram.

Three product rules matter for operators:

  1. Read is per user. Marking an alert read hides it from your dashboard unread list. Teammates still see it unread until they mark it.
  2. Delete is admin-only. Members cannot wipe the project history.
  3. Retention follows the plan of the project’s first admin: 7 days (Free / no active sub), 90 days (Pro), 365 days (Pro+). Exact copy is on Pricing.

How the inbox behaves

SurfaceUnreadRead alerts
Dashboard homeRecent unread (warning/critical still arrived on channels)Hidden after you mark read
Alerts searchAll matching rowsShown with a read badge
Chat destinationsIndependentIndependent

Info alerts never hit channels (levels) but they still occupy retention until they age out.

Setup (once)

  1. Open app.thealerthub.com on a project you belong to.
  2. Confirm your role (Admin vs Member) under project members.
  3. Send a test warning, mark it read, then search for it in Alerts.

No extra webhook is required. This is the app, not a source integration.

Workflow 1: On-call inbox

Treat the dashboard like email unread:

  • Morning: unread only — leftover criticals from the night.
  • After you refunded the Stripe customer or restarted the job, mark as read.
  • Leave info-level Shopify volume in search; do not expect it on the home unread list if you marked as you went (or if you never paged — info is easy to ignore on home depending on filters).

Workflow 2: Admin cleanup

Wrong test payloads, leaked keys in a title, or a looped webhook: an admin deletes the row. Members keep read/unread only. Deleted alerts drop related read receipts.

Do not use delete as “archive.” Retention already expires old rows. Delete is for mistakes and abuse.

Workflow 3: Choose a plan for history

Investigating a dunning incident from four months ago needs Pro+ (365 days) or an export habit before expiry. Free is a rolling week — enough to debug last night’s deploy, not last quarter’s churn.

Upgrade on Pricing if the feed is your system of record.

Ideas to squeeze more value

  • Do not duplicate archives in Google Sheets. Search the feed; raise retention if you outgrow 7 days.
  • Agencies. One project per client so retention and unread are not a mixed bag of shops.
  • WhatsApp is not the archive. Quota and disappearing chats are why the feed exists. WhatsApp Pro.

FAQ

Can I unmark as read?
Not in the current UI. Search still lists the alert with a read badge.

Does marking read stop Telegram?
No. Delivery already happened. Read is only the AlertHub inbox.

Who owns retention on a shared project?
The first admin (by join time) and that admin’s subscription limits.

Next steps

Mark last night’s leftover warnings read, check your plan’s retention on Pricing, and keep sources pointed at the same project.

Related: alert levels · mobile push