Auditing Warehouse Source Health

Auditing data warehouse source health

This skill produces a project-wide audit of the source and sync side of the data warehouse pipeline — source connections, sync schemas, and webhook push channels. Use it when the user wants a summary of what’s broken with their imports, not a deep-dive on one sync. The deep-dive on individual failures is diagnosing-failed-warehouse-syncs; this skill is the scan that tells them where to look first.

The same underlying endpoint (data-warehouse-data-health-issues-retrieve) also reports materialized-view, batch-export-destination, and transformation issues. Materialized views are covered by auditing-warehouse-view-health. Destinations (batch exports) and transformations are owned by other products — surface them if they appear, but route them to the relevant team rather than diagnosing here.

When to use this skill

• “Why are my imports failing?” / “What’s broken with my sources?”
• “Why is my warehouse data stale?”
• The user is new to a project and wants to know which sources they’ve inherited and whether they’re healthy
• Weekly or monthly review of source/sync health
• Dashboards are stale and the user isn’t sure which source is at fault

Available tools

The data-health-issues endpoint aggregates across the whole pipeline — it’s the fastest path to a summary. Filter its results to the source and external_data_sync types for this audit. Use the list endpoints when you need more context than the summary provides (row counts, non-failing items, schema-level detail).

What counts as a source/sync “issue”

From the data-health endpoint, this audit cares about two of the five categories:

Each entry includes id, name, type, status, error, failed_at, url, and source_type.

The other categories the endpoint returns are out of scope for this skill:

• materialized_view → auditing-warehouse-view-health
• destination (batch export) → owned by the batch exports / data pipelines product
• transformation (HogFunction) → owned by the CDP / ingestion side

Note the data-health endpoint only reports active failures. For source/sync health it doesn’t flag:

• Schemas paused by the user (should_sync = false)
• Schemas that are slow or stale but technically Completed
• Webhook problems on sync_type: "webhook" schemas. The bulk-sync safety net can succeed while the webhook push channel is silently broken (deregistered, disabled on the remote side, failing signature verification). These don’t surface in data-health-issues — check per-source with webhook-info-retrieve.

If the user asks about staleness or unused items, reach beyond this endpoint — see Step 4.

Workflow

Step 1 — One-shot pull

Call data-warehouse-data-health-issues-retrieve and keep the source and external_data_sync entries.

If there are no source/sync issues, tell the user their sources are healthy and stop. Don’t invent problems.

Step 2 — Group and prioritize

1. Sources in Error first. A source failure cascades — every schema under it is effectively dead until the source reconnects. Fix these first.
2. Sync schemas next, in this order:
   • status: "billing_limit" entries (billing issue, non-technical — flag and route to billing)
   • Failed on heavily-used tables (user asks / check row counts via schemas-list if needed)
   • Failed on less-used tables

Step 3 — Present the audit

Render a prioritized report. Don’t dump the raw JSON — human-readable table per category:

## Data warehouse source health — 4 issues

### 🔴 Sources (1)
- Stripe — authentication failed (failed 2h ago). All 8 tables under it are currently dead.
  → `diagnosing-failed-warehouse-syncs` on this source

### 🟠 Sync schemas (3)
- postgres_prod.orders (Failed 6h ago) — column "updated_at" does not exist
- postgres_prod.invoices (Failed 6h ago) — column "updated_at" does not exist
- hubspot.contacts (BillingLimitReached) — team quota exceeded

Recommended order:
1. Stripe auth (everything under it is dead)
2. Schema-drift on postgres_prod.orders / invoices — looks like upstream renamed a column
3. Billing limit on hubspot

The exact format is less important than: prioritized, grouped, actionable, and hinting at the right next skill.

Step 4 — Go beyond active failures (when asked)

If the user wants more than just “what’s on fire” — e.g. “what else should I look at?” — cross-check:

Stale but “Completed” schemas: Call external-data-schemas-list and look for schemas with old last_synced_at relative to their sync_frequency. A schema on 1hour frequency that last synced 3 days ago is effectively broken even if status says Completed.

Sources with zero sync activity: Sources where every schema has should_sync: false or status = Paused. These were set up and then abandoned — candidates for cleanup via external-data-sources-destroy.

Broken webhooks on webhook-type schemas: Iterate the sources that have any schema with sync_type: "webhook" (visible via external-data-schemas-list). For each, call external-data-sources-webhook-info-retrieve({source_id}):

• exists: false while a schema is sync_type: "webhook" → webhook was never registered, or was deleted. Push channel is dead; only the bulk fallback is ingesting.
• external_status.error present → remote service is reporting a problem (permission revoked, endpoint deleted on their dashboard).
• external_status.status not "enabled" → remote has disabled the endpoint (often after repeated delivery failures).

Report these separately from the primary audit — they’re a different shape of problem than failed syncs, and the fix is a different skill (diagnosing-failed-warehouse-syncs scenario I, or setting-up-a-data-warehouse-source step 5.5).

Only run these extra checks if the user explicitly asks for a broader audit — they involve more tool calls and heuristics.

Step 5 — Offer the next step

End the audit with a clear hand-off:

• “Want me to dig into the Stripe failure?” → hands off to diagnosing-failed-warehouse-syncs
• “Want me to fix the schema drift on orders?” → hands off to tuning-incremental-sync-config
• “Want to disable the billing-capped schemas?” → one-click via external-data-schemas-partial-update

Never start applying fixes autonomously from an audit — the audit’s job is to report and recommend, not remediate. Any fix should be confirmed explicitly before executing.

Important notes

• The audit is read-only. Never call destructive tools from the audit flow. Hand off to the diagnosis/tuning skills — which in turn confirm before acting.
• Empty = healthy. Don’t pad an empty audit with hypothetical issues. “No source issues found” is a good answer.
• Source failures cascade. When reporting a source in Error, also mention which schemas under it are affected (or will be, once they try to sync again). The user needs to understand the blast radius.
• Billing limits aren’t technical problems. Flag them but route to billing / quota discussion, not to a recovery action.
• data-health-issues only surfaces active failures. For staleness or abandoned sources you need to cross-check the list endpoints. Only do this when the user explicitly asks for a deeper audit.
• Webhook health is separate from schema health. The data-health endpoint doesn’t know about webhook state. When a user’s request mentions “real-time”, “Stripe webhook”, or “why is data hours behind on a webhook source”, go straight to webhook-info-retrieve rather than inferring from schema status.
• Materialized views, destinations, and transformations are out of scope here. They share the data-health endpoint but belong to other audits/products — route, don’t diagnose.