| name | rca-diagnostics |
| description | Cross-system incident response and post-mortem Root Cause Analysis for data platforms.
Use when: investigating MULTI-system failures, correlating failures across infra + data + code, writing post-mortem reports, or diagnosing cascading failures.
Do NOT use for single Airflow DAG failures — use `debugging-dags` instead.
Do NOT use for single dbt model errors — use `debugging-dbt-errors` instead.
|
| license | MIT |
| metadata | {"author":"dataops-agent","version":"2.0.0"} |
Root Cause Analysis (RCA) Diagnostics
Cross-system incident response. For single Airflow DAG failures, use debugging-dags. For single dbt errors, use debugging-dbt-errors.
When to Apply
Use this skill when:
- Multiple pipelines or systems failed simultaneously (cascading failure)
- Correlating failures with infrastructure changes (deploys, config changes)
- Correlating failures with upstream data source changes
- Writing post-mortem / incident reports
- Investigating systemic reliability issues (recurring failures, intermittent errors)
- Analyzing patterns across multiple incident occurrences
Scope Boundaries
| Task | Correct Skill |
|---|
| "My DAG failed, fix it" | debugging-dags |
| "dbt build failed with error" | debugging-dbt-errors |
| "3 pipelines broke after deploy" | this skill ✅ |
| "Write a post-mortem for the outage" | this skill ✅ |
| "Why do these errors keep recurring?" | this skill ✅ |
Best Practices
- Methodical Tracing: Work backward from the point of failure to the source.
- Holistic View: Look beyond the code (check infrastructure, upstream API changes, network timeouts).
- Blameless Culture: Focus on systemic failures and how to prevent them in the future.
- Actionable Fixes: Always provide a short-term mitigation and a long-term architectural fix.
Execution Pattern
- Gather all available logs, error messages, and context from all affected systems.
- Build a timeline of events leading to the incident.
- Perform backward tracing to isolate the root cause.
- Provide a clear explanation of why it failed.
- Propose immediate remediation and long-term preventative measures.
- Draft a post-mortem document if requested.
