← ClaudeAtlas

firewatchlisted

Investigate production alerts - assess severity, search logs, trace errors, and create JIRA tickets
sgtlaugh/skills · ★ 2 · AI & Automation · score 76
Install: claude install-skill sgtlaugh/skills
# /firewatch - Production Alert Investigation Investigate production alerts end-to-end: assess severity, match known patterns, search logs, trace through code, and optionally create a JIRA ticket. ## Usage ``` /firewatch <paste alert text or description> /firewatch 500s spike on <service> /firewatch Opsgenie alert #<tinyId> /firewatch check all services ``` ## Service catalog Read `~/.config/claude-skills/services.md` **before Step 1** - every step depends on it. Its sections: - **Global** - GitHub org, clones dir, *Log search skill* (`<log-skill>` below), spread tags - **Services** - per service: repo, Sentry project(s) (`none` = no Sentry project), log service, notes on service boundaries - **Alerts** - incident channel, on-call guide, Jira project + label for alert tickets, typical cascade, and known patterns specific to these systems - **Jenkins**, **Datadog**, **Opsgenie** - base URL and job naming, env tag and service-to-APM-metric table, alert-stream notes - **Health check** - the queries for `check all services` Assessing severity or matching patterns without it is WRONG - known patterns and service boundaries live there. A value the catalog lacks is asked for or marked inferred, never guessed. *Log search skill* `none` → log steps are `Can't-determine: no log skill configured`. ## Process ### Step 1 - Parse Alert & Assess Severity Extract from the alert text: - **Service name** - matched against the catalog's Services table - **Error type** (500s, timeout, f