The release quality loop
A validation result becomes useful when a team can compare it, investigate it, fix it, and reproduce the outcome.
One loop, six stages.
Use the chapters to move from the accepted basis through validation, comparison, investigation, decision, and preserved evidence.
The operating model stays stable across every workflow.
Use cases change the trigger and decision. These six stages keep the measurement and evidence comparable.
- Step 1 of 6.
Basis
Accept the measurement basis
Pin the server, release, scope, profiles, terminology, rules, and thresholds before the run.
- Step 2 of 6.
Validate
Measure the release scope
Run the same deterministic validation contract against the selected environment or dataset.
- Step 3 of 6.
Compare
Read the delta
Compare the current result with the accepted run instead of treating every known finding as new.
- Step 4 of 6.
Investigate
Trace the signal
Move from the gate to the run, pre-filtered issues, affected resource, field path, and constraint.
- Step 5 of 6.
Decide
Re-run the remediation
Verify the changed scope and let the team make the promotion, acceptance, or handover decision.
- Step 6 of 6.
Evidence
Preserve the next basis
Keep the result, configuration fingerprint, package pins, timestamps, and integrity metadata together.
Where the loop shows up
The same six stages, applied to a specific decision.
Ship safely
Release Safety Gate
Run the loop before a promotion and decide from the delta, not from the absence of incidents.
Understand change
Drift & Regression Detection
Run it on a schedule and separate a real regression from pre-existing debt.
Ship safely
CI/CD Quality Gate
Run it in the pipeline so the exit code carries the decision.
Prove readiness
Regulatory & Audit Evidence
Keep the last step — the exported evidence — as the artifact an auditor receives.
Walk the loop on a real server.
Browse a public FHIR server and see the same surfaces without an account.