Early accessBy invitation, and the SDKs aren't published yet.Ask for an invitation
Skip to Content
Docs

Alerts

Emails when a new kind of error appears, when one spikes, or when a version starts failing.

On Pro and Business, a project can email you when something breaks, and post to Slack or (on Business) to a webhook of yours. Nobody gets alerts they didn't ask for: each person chooses theirs in the project's settings, under Alerts. Free organizations have no alerts.

What can email you

AlertWhen
A new kind of errorAn error the project hasn't had in the last 30 days: a new area and name. Once per error.
An error spikesIn the last hour, one error happened at least 20 times and at least 3 times its hourly average over the week before. At most once every six hours per error.
A version starts failingThe newest version with at least 50 sessions has 5 points fewer error-free sessions than the version before it. Once per version.

Errors are events at the error and critical levels. Alerts watch the environment the dashboard opens on (see Environments), so development builds don't wake anyone. They're checked every five minutes; an event that arrives late is still noticed.

Each email says which error or version, how many events and people, and links to the project's Errors or Releases page. It never carries more of an event than the dashboard's list shows.

Choosing yours

In Alerts, under Email me, turn on the ones you want. They go to your account's email, which has to be confirmed. Anyone in the project can choose theirs; leaving the organization, or being removed from it, stops them.

For everyone in the project

Owners choose which alerts the project sends at all, and when a spike or a version counts:

SettingDefaultRange
A spike is this many times the usual rate31.5 to 20
With at least this many events in the hour201 to 100,000
A version is failing at this many points fewer error-free sessions51 to 50
Once it has this many sessions5010 to 100,000

An alert an owner turns off stops for everyone in the project, whatever they chose.

Plans

PlanAlerts
FreeNone
ProEmail and Slack
BusinessEmail, Slack and webhooks

Slack

Owners connect one Slack channel per project, under Also send to:

  1. In Slack, create an incoming webhook for the channel (an app with Incoming Webhooks turned on, then Add New Webhook).
  2. Paste its address (it starts with https://hooks.slack.com/services/) in Slack, choose which alerts go there, and choose Connect.
  3. Send a test posts a message to check it arrives.

The address is a secret: anyone who has it can post to the channel. Lyrelog never shows it again whole, only its end, and never logs it. Messages are posted for everyone in the project, whatever each person chose for email. If Slack answers that the webhook is gone, Lyrelog stops trying; remove it and connect a new one.

Webhooks

On Business, owners can also connect one webhook: any https:// address on port 443 that Lyrelog can reach on the internet. Each alert is a POST with a JSON body:

{
  "type": "alert",
  "kind": "new_error",
  "project": { "id": "a83810a9e984", "name": "Tilefall" },
  "title": "A new kind of error in Tilefall",
  "detail": "Tilefall has an error it hadn't had in the last 30 days.",
  "url": "https://app.lyrelog.com/p/a83810a9e984/errors",
  "issue": { "area": "save", "name": "Couldn't write the save file", "events": 4, "people": 3 },
  "release": null,
  "sent_at": "2026-11-20T12:54:36Z"
}

kind is new_error, spike, release or test. A spike carries issue with the last hour's events and people; a failing version carries release instead:

"release": { "version": "2.1", "previous": "2.0", "error_free": 0.8, "previous_error_free": 0.95, "sessions": 200 }

Checking it came from Lyrelog

When you connect the webhook, Lyrelog shows its signing secret once (whsec_...). Each request carries X-Lyrelog-Signature: t=<unix seconds>,v1=<hex>, where v1 is the HMAC-SHA256 of <t>.<body> with that secret. Recompute it over the raw body, compare in constant time, and refuse a t more than five minutes old:

import { createHmac, timingSafeEqual } from "node:crypto"

function fromLyrelog(rawBody, header, secret) {
  const { t, v1 } = Object.fromEntries(header.split(",").map((part) => part.split("=")))
  if (Math.abs(Date.now() / 1000 - Number(t)) > 300) return false
  const expected = createHmac("sha256", secret).update(`${t}.${rawBody}`).digest("hex")
  return v1.length === expected.length && timingSafeEqual(Buffer.from(v1), Buffer.from(expected))
}

Delivery

Answer with any 2xx within 10 seconds. Lyrelog doesn't follow redirects. A 5xx, a 408, a 429 or no answer is tried again later (30 seconds, then doubling, up to 8 tries); any other 4xx is given up. The settings show how the last delivery went.