Notifications

Channels

Telegram · Slack · Discord · Webhook · Email

Events

12 toggles per channel

Scope

Platform-wide, owner/admin

Flow

1. 01eventdeploy, backup, job, container down, disk, certificate, DNS, log errors
2. 02filterchannels with that event's toggle enabled
3. 03formata per-platform payload: HTML, attachments, embeds, JSON, email
4. 04sendfetch with a 10-second timeout; failures are only logged

Supported channels

Adding a channel

1. Integrations → Notifications → New notification channel (owner/admin)

   Fill in a Name, pick a Type, fill in its fields, and enable the toggles for the events you want. The token/URL/password is stored encrypted and never shown again — the channel list only shows a hint of the target (e.g. chat id, domain).

2. Send a test

   The test button sends a sample message to that channel. Provider errors are shown as-is — e.g. Telegram's chat not found (the bot isn't in the group / wrong chat id) or SMTP's 535 Authentication failed.

3. Trigger a real event

   Deploy once, or run a job that's meant to fail (exit 1) to make sure the toggle works.

Events

• Deploy started is off by default (unlike the other events) — deploys can happen often (auto-update, webhooks), so a channel has to opt in if it actually wants to know. It also never fires for a kind: config deployment (a swarm mode/ replica/resource-limit change on the same image) — that isn't a new deploy worth watching for.
• A link to the application page is included when WEB_ORIGIN is set.
• A failure to deliver to one channel is only logged by the API and doesn't affect other channels or the process that triggered it (a deploy never fails because of a notification).
• Container down: at most 1 notification per container every 10 minutes.
• Remote server unreachable and HTTP monitor failing are on by default, but only do anything when you use the feature (adding a remote server / turning on a monitor for an application), so they are not noisy for anyone who doesn't. Each uses a single toggle for the failure and the recovery message (told apart by the level and by event in the webhook payload: server.down/server.recovered, http.down/http.recovered).
• DNS domain issue: two consecutive failed checks (to dampen transient DNS blips) before the first notification, then at most once a day per domain while the status stays bad. A domain with nothing to compare against (no expected IP) never sends.
• Application error is off by default — detection from plain-text logs is prone to false positives. At most 1 notification per application every 15 minutes, except an error shape never seen since the API last started, which is always sent right away. Can be disabled per application with the Ignore error logs switch on the application's Settings tab — see Applications.

Generic webhook payload

The Webhook channel sends a POST with Content-Type: application/json. Common fields: title, level (success / failure / info), url, plus event-specific fields:

{
  "title": "Deployment started: shop",
  "level": "info",
  "url": "https://panel.example.com/applications/…",
  "event": "deployment.started",
  "deploymentId": "…",
  "kind": "build",
  "applicationId": "…",
  "application": "shop",
  "project": "Toko",
  "trigger": "webhook",
  "commitSha": "a1b2c3d4e5f6",
  "commitMessage": "fix: checkout bug",
  "triggeredBy": "octocat"
}

{
  "title": "Deployment succeeded: shop",
  "level": "success",
  "url": "https://panel.example.com/applications/…",
  "event": "deployment.success",
  "deploymentId": "…",
  "kind": "deploy",
  "applicationId": "…",
  "application": "shop",
  "imageRef": "localhost:5000/toko/shop:a1b2c3d4e5f6",
  "error": null,
  "finishedAt": "2026-09-22T04:30:00.000Z"
}

{
  "title": "Container down: aoox-app-shop",
  "level": "failure",
  "event": "container.down",
  "container": "aoox-app-shop",
  "exitCode": 137,
  "restarting": true
}

{
  "title": "Application error detected: shop",
  "level": "failure",
  "url": "https://panel.example.com/applications/…",
  "event": "app.error",
  "applicationId": "…",
  "count": 2,
  "fingerprints": ["TypeError: Cannot read <n>", "FATAL: <n>"]
}

Example receiver (Node.js)

app.post("/hooks/aoox", express.json(), (req, res) => {
  const { event, level, title } = req.body
  if (event === "deployment.failure") alertOnCall(title)
  res.sendStatus(204)
})

Headers & signature

const sig = req.get("X-Aoox-Signature")
const mine = "sha256=" + createHmac("sha256", SECRET).update(rawBody).digest("hex")
if (!sig || !timingSafeEqual(Buffer.from(sig), Buffer.from(mine))) return res.sendStatus(401)

Common pitfalls

Next steps