What problem does it solve? Failed Cloud Composer (Apache Airflow) pipelines are hard to debug because logs, remote DAG code, and environment configuration are scattered across GCP services. This Skill guides an evidence-based Root Cause Analysis (RCA) workflow that correlates logs with the actual code running in the Composer environment. ## Core Features & Use Cases - Log-Based Evidence Gathering: Uses gcloud logging read with filters on severity, environment name, and DAG ID to locate failures and stack traces. - Remote Code Retrieval: Downloads the actual DAG code from the environment's GCS bucket to ensure analysis matches what is really running, not stale local copies. - Structured RCA Reports: Pinpoints the exact failing line or configuration, proposes a fix, and generates a saved RCA report. - Pipeline Template Validation: Provides compliant templates for declarative deployment.yaml and orchestration-pipeline.yaml files covering dbt, PySpark, and BigQuery actions. - Use Case: A user's daily_sales_agg DAG failed yesterday; the Skill fetches error logs, downloads the remote DAG, identifies a KeyError on line 45, and produces an RCA report with a recommended fix. ## Quick Start Ask the assistant to troubleshoot a failed DAG in your Composer environment by providing the DAG ID and approximate failure time, and request a root cause analysis report.
