Skills Data and integration

Monitoring support data quality

The failure this exists to catch: a sync stops, and every dashboard keeps rendering. Charts still draw, averages still compute, and the only symptom is that the last few days look quiet — which is indistinguishable from a quiet few days.

Data and integrationAnalysisAny helpdeskRead-only
Installnpx rulebase-skills install cx-data-quality-monitoring

When to use it

Reach for this when someone says any of these — they are the phrases the skill itself triggers on:

  • our sync stopped and nobody noticed
  • monitor support data quality
  • a channel stopped syncing
  • the dashboard looked fine but the data was stale

How it works

The method, in the order the skill runs it. The full procedure — tables, worked examples and the edge cases — is in the skill itself.

  1. Monitor per segment, not in aggregate

    This is the single most important design decision. A total volume check will not catch one channel of six going dark, because the dip sits inside normal variation.

  2. The checks, in order of what they catch

    1. Freshness, per segment. Time since the most recent record. The cheapest and highest- yield check.

  3. Distinguish "no data" from "no activity"

    The core ambiguity, and the reason freshness alone is not enough.

  4. Authorisation expiry, which is the most common real cause

    OAuth tokens expire, credentials get rotated, an admin removes an integration user, a scope gets narrowed. The result is a sync that runs, returns fewer records or none, and logs a 401 somewhere nobody reads.

  5. Make the alerts survivable

    An alert nobody reads is a monitor that does not exist: Route to an owner, not a channel, Suppress during known maintenance, and alert if maintenance runs long, Review the alerts periodically and retire the ones nobody acts on.

  6. Surface data health where the data is consumed

    The most valuable output is not the alert. It is that a dashboard shows when its data was last complete, so the reader can tell. A freshness and completeness indicator on the report itself prevents the specific failure this skill is about — someone acting on partial data without knowing it is partial.

Related skills

Free and open source, and vendor-neutral — it reads the conversations from whichever helpdesk you already run. Browse all 149 skills · connect your helpdesk over MCP · source on GitHub

Review every conversation. Act on what it finds.

AI for customer operations, built for financial services. Specialist agents chase every issue to resolution and every stalled customer to activation.

Rulebase dashboard