Incident Triage Alert

Receive a Sentry issue webhook, parse the real issue payload, summarize it with AI, and notify Telegram contacts tagged admin through the installed bot audience.

The user wants an internal operations assistant that reacts to Sentry-compatible issue webhooks and alerts admins in Telegram with a concise AI triage summary.

• Should the admin audience use the default admin tag or a different existing tag?
• Which Sentry issue actions should alert admins?
• Should the first version only notify, or also create a GitHub/Jira issue later?
• In Sentry, add a webhook integration that posts issue events to the bot webhook URL ending in /webhooks/sentry/issues.
• Tag internal admin contacts in the installed bot audience with admin, or change adminAudienceTag to the existing admin tag.
• Send a Sentry test issue webhook and confirm admins receive the parsed title, level, project, URL, and AI summary.
• Review the AI prompt so it matches the team's incident severity policy and escalation rules.

• Sentry sends an issue webhook to /webhooks/sentry/issues
• Webhook output fields parse data.issue fields into variables
• The flow runs in broadcast mode for contacts tagged admin
• AI_ACTION creates a compact incident summary from the parsed Sentry payload
• A Telegram message is sent by the installed bot to every matched admin contact
• Admins can extend the flow later with acknowledgement or escalation steps

Good to know

• This template is built for the Sentry-compatible issue webhook shape: action and data.issue.*.
• No Telegram chat ID is hardcoded. The installed bot supplies eligible contacts, and the broadcast filter limits delivery to contacts tagged admin.
• After bot install, tag internal admin contacts with admin or change adminAudienceTag to the team's existing admin tag name.
• The webhook intentionally has no contactId mapping. Contactless webhook execution is allowed because the flow is audience-scoped by its broadcast filter.
• Keep automated remediation out of the first version. Start with triage, notification, acknowledgement, and handoff.

Keep in mind

• Do not notify customer audiences or broad broadcasts with internal incident details.
• Do not hardcode personal Telegram IDs or a group chat ID for the default path.
• Do not claim the incident was fixed or deployed unless a real remediation integration exists.
• Do not expose secrets, tokens, stack traces, or private payload fields outside the admin audience.