Skills Quality assurance
Auto-fail governance
Auto-fail criteria exist because some failures cannot be averaged into a 92%. The second failure mode is recording them and doing nothing — which transforms a quality programme into a documented compliance gap.
npx rulebase-skills install cx-auto-fail-governanceWhen to use it
Reach for this when someone says any of these — they are the phrases the skill itself triggers on:
- “auto-fails nobody actions”
- “QA compliance failures sitting open”
- “hard fail vs soft fail”
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.
Step 1: Separate hard fails from soft fails
These must not share a workflow or a name in reporting.
Step 2: Every auto-fail needs five fields at creation
Non-negotiable on the evaluation record: Criterion id and version, Evidence, Severity tier, Named owner, SLA clock.
Step 3: Define SLAs by tier
Set SLAs the organisation can actually meet — aspirational SLAs that are missed daily are worse than no SLA.
Step 4: Build the evidence trail
Each closure needs linked artefacts: Initial evaluation with evidence, Review notes, Remediation, Sign-off.
Step 5: Route and notify correctly
Avoid duplicate tickets across QA tool and helpdesk with no link. One id should reference the other.
Step 6: Audit the unactioned backlog
Run monthly (minimum).
Step 7: Connect to outcomes, not only volume
Governance metrics that matter: Recurrence rate, Customer remediation rate on hard fails where customer harm possible, Criterion false-positive rate, Time-to-close trend.
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
