> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scaling.cloud/llms.txt
> Use this file to discover all available pages before exploring further.

# Sentry

> Send Sentry alerts to scaling.cloud to page your on-call, and let AI investigations read Sentry issues, releases, and stack traces for root-cause findings.

The Sentry integration connects your Sentry organization to scaling.cloud in **two directions** from a single install:

* **Send alerts → page on-call.** A Sentry issue-alert rule fires, scaling.cloud opens an [incident](/concepts/incidents) and pages whoever is on-call through your [escalation policy](/concepts/escalation-policies).
* **Read for investigations.** [AI investigations](/concepts/ai-investigations) read your Sentry issues, releases, and stack traces and surface them as root-cause findings.

You install once. Both directions are authorized by the same connection — there are no per-project tokens or webhook URLs to copy.

## Send Sentry alerts as incidents

Add **"Send a notification via scaling.cloud"** as an action on any Sentry issue-alert rule. When the rule fires:

* scaling.cloud opens an incident and routes it through your org's [routing](/concepts/routing-policies) and [escalation policy](/concepts/escalation-policies), paging the engineer who is on-call right now.
* The Sentry event level maps to incident severity — `fatal` → critical, `error` → high, `warning` → medium, `info`/`debug` → low.
* Alerts are **deduplicated by Sentry issue**, so a noisy rule raises one incident, not hundreds.
* When the Sentry issue is **resolved**, the matching incident resolves automatically.
* The incident maps to the [component](/concepts/components) whose name matches the Sentry **project slug**; if none matches, a component-less incident is opened so the page still happens.

<Card title="Send Sentry alerts to on-call" icon="bell" href="/guides/send-sentry-alerts">
  Step-by-step: install, add the alert action, and verify a page.
</Card>

<Note>
  Auto-created incidents are never auto-listed on your public [status page](/concepts/status-pages) — promote one manually if it warrants customer-facing communication.
</Note>

## Enrich investigations (read)

The same connection is a [signal source](/concepts/signal-sources) for [AI investigations](/concepts/ai-investigations). Once connected, an investigation can read your Sentry data and surface findings:

* `root_cause` — derived from a grouped error's stack trace.
* `commit_correlation` — from a release's linked commits, corroborating a GitHub [suspect change](/integrations/github) at commit precision.

This direction is **read-only** and enriches investigations on existing incidents — it does not open them.

### Watched projects

After connecting, you choose which Sentry projects investigations may read from, under **Watched projects**. Only the projects you select are read — connect once, then scope the integration to the projects that matter for your incidents.

## Connect

Connecting authorizes scaling.cloud against your Sentry organization through Sentry's integration platform. You need the **Admin** role in both scaling.cloud and Sentry.

<Card title="Connect Sentry" icon="bug" href="/guides/connect-sentry">
  Step-by-step authorization and project selection.
</Card>

<Note>
  scaling.cloud stores the Sentry access and refresh tokens encrypted and refreshes
  them automatically — you connect once and both directions keep working across token
  lifetimes.
</Note>

## Remove the integration

Open **Settings → Integrations → Sentry** and click **Disconnect Sentry**. scaling.cloud stops reading from your Sentry organization and stops accepting its alerts immediately.