What problem does it solve? When checks on a DeepSource-style analysis platform time out, the database stores no error payload because the watchdog kills the pod before it reports, leaving engineers blind to root causes. This Skill recovers the missing evidence by pulling raw analyzer pod logs and backend timeout verdicts from GCP Cloud Logging, classifying each failure, and presenting everything in an interactive local dashboard. ## Core Features & Use Cases - End-to-end timeout investigation: Queries the analysis-DB read replica via the mb CLI, fetches pod logs and watchdog verdicts via gcloud logging read, and classifies root causes such as clone failures, lost results, OOM kills, slow scans, and infra overhead. - Interactive HTML dashboard: Generates a self-contained index.html with daily trend charts, root-cause and repository filters, search, per-check lifecycle timelines, and links to raw logs and DB payloads. - Use Case: Ask why the secrets analyzer timed out between July and September; the Skill produces a dashboard showing that most failures were clone failures on deleted branches rather than slow scans, letting you route fixes to the right team. ## Quick Start Ask the AI to analyze why the secrets analyzer timed out between 2026-07-01 and 2026-09-01 and generate the timeout dashboard in a local output directory.
