ProductAlerts

Alerts

In-dashboard alert rail, velocity spikes, regressions, ingest cap signals, and how they relate to webhooks and snooze.

Alerts surface actionable signals without opening every issue row. Open the Alerts rail in the project shell (alongside Issues). Webhooks can mirror the same events to Slack or Discord.

Alerts are not a separate product from Issues — they notify you when issue lifecycle rules fire.

What can fire an alert?

Built-in kinds (filter chips in the UI):

KindTypical meaning
new_issueFirst event for a fingerprint
velocity_spikeEvent rate jumped in a rolling window (default 15 minutes, ratio ≥ 4× vs prior window)
regressionResolved-in-release issue came back on a newer release
event_milestoneOccurrence count crossed a rule threshold
users_affectedUnique users crossed a rule threshold
ingest_cap_hitProject hourly ingest cap reached (403 on further ingest)
customCustom alert rules you define in the rules panel

Tune velocity window with EPURE_VELOCITY_WINDOW_SECS on the server: Configuration.

Spike valve ≠ velocity alert

The spike valve (~100 events/min/fingerprint) still returns 202 and may drop duplicate bodies. Velocity alerts compare windows over minutes and show in the Alerts rail — different mechanism: Concepts.

Alerts vs webhooks vs snooze

MechanismPurpose
Alerts railIn-app unread/history, mark read, jump to issue
WebhooksOutbound HTTPS to Slack, Discord, or generic JSON
SnoozeMute an issue (time, occurrences, or users) — does not delete alerts history

Webhook event ids include issue_created, regression, and velocity_spike. Setup: Webhooks API and Settings.

Velocity spike (operator mental model)

  1. Epure tracks timestamps per issue in memory.
  2. When the current window’s count vs the previous window exceeds the ratio threshold, it inserts a velocity_spike alert (and may enqueue webhook dispatch when subscribed).
  3. Use it to catch deploy regressions and traffic floods before manual triage.

Regression alerts tie to resolve in release workflow: Releases.

Ingest cap alerts

When a project exceeds ingest_cap_per_hour (default 5000), further ingest returns 403 ingest_cap_exceeded. Epure may also record an ingest_cap_hit alert so operators see saturation in the rail — not only failed SDK posts.

Raise the cap or fix the logging loop: Projects, API errors.

Agent API

Automation can list alerts with a personal access token: GET /api/v1/agent/alerts. See Agent API.

Why did I get a velocity alert but Issues looks quiet?

The alert fired on rate change for one fingerprint. Snoozed or ignored issues still generate signals. Open the alert row to jump to the issue and check environment filter and time window.

Can I pause a rule?

Use the rules panel in Alerts to pause built-in or custom rules per project. Ignored alerts move to the Ignored view without deleting history.

Do alerts email me?

Phase 1 ships in-app rail + webhooks. Point a generic webhook at your mail bridge if you need email.