triage

name	triage
description	Run SWEny triage to investigate production alerts, create issues, and optionally implement fixes. Use when the user wants to analyze recent errors or alerts.
disable-model-invocation	true
allowed-tools	Bash
context	fork
agent	sweny-workflow
argument-hint	["--dry-run"]

SWEny Triage

Run the SWEny triage workflow. This investigates production alerts from configured observability providers (Datadog, Sentry, BetterStack, etc.), performs root cause analysis, creates issues, and optionally implements fixes.

Prerequisites: A .sweny.yml config file and provider credentials must be set up. Run /sweny:check first if unsure.

Usage

Run triage with default settings:

sweny triage --stream

Run in dry-run mode (investigate only, skip creating issues/PRs):

sweny triage --stream --dry-run

If the user provided arguments, pass them through:

sweny triage --stream $ARGUMENTS

What to expect

• The workflow takes 2-10 minutes depending on alert volume
• Progress events stream as NDJSON (node:enter, node:progress, node:exit)
• On completion, report a concise summary: how many alerts found, issues created, PRs opened
• If it fails, check sweny check output for credential issues

Common flags

• --dry-run — investigate without creating issues or PRs
• --time-range <range> — time window to analyze (default: 24h)
• --severity-focus <focus> — severity level (default: errors)
• --service-filter <filter> — filter by service pattern