| name | dalgo-troubleshooting |
| description | Diagnose Dalgo failures — failed pipeline runs, sync errors, dbt errors — and handle notifications. Use when a user asks why something failed, what went wrong, or what their notifications say. |
dalgo-troubleshooting
Use for failure diagnosis and notifications through Dalgo MCP tools.
Always
- Diagnose from evidence: fetch history and logs before naming a cause. Quote the actual error line, then explain it in plain language.
- Work backwards through the data flow: visualization → warehouse → transforms (dbt) → pipeline run → sync → source.
- End every diagnosis with a concrete suggested action the user can take.
- For "how does Dalgo handle X" questions during diagnosis, check
dalgo_search_docs before answering from memory.
- Do not re-trigger a failed run as a diagnostic step; understand the failure first, then offer the re-run.
Decision Rules
- "Why did pipeline X fail?":
dalgo_get_pipeline_run_history → failed run's flow_run_id → dalgo_get_flow_run_logs → quote and translate the error.
- "Why is my data stale/missing?": check the last sync (
dalgo_get_sync_history), then the last pipeline run, then whether dbt models built (dalgo-transforms).
- dbt errors in pipeline logs: use the error translations in
dalgo-transforms.
- Sync errors: use the triage patterns in
dalgo-ingestion.
- "Any notifications?" / "what did I miss?":
dalgo_get_unread_count, then dalgo_list_notifications; only dalgo_mark_notifications_read when the user asks.
Workflow Order
- Locate the failing object and its most recent failed run.
- Fetch logs or sync history for that run.
- Identify the error line; classify it (credentials, schema drift, network, dbt model, permissions).
- Explain cause and remedy in plain language; offer the fix or re-run.
