Green pipeline.
Broken release.
A controlled GNU Group demonstration showing how a release can appear successful while delivering incorrect business behaviour—and what better release controls look like.
One deliberately defective calculation passed into a local simulated deployment because its test failure was ignored.
The same defective version was blocked. Passing versions were identified by content digests, checked after promotion and restored by prior digest.
What GNU Group evaluated
Build and quality gates
Does a failed test stop promotion, or is failure bypassed in the pipeline?
Release identity
Does deployment use an immutable, independently verified artifact?
Recovery confidence
Can a previous known-good artifact be restored using recorded evidence?
Evidence-backed example findings
Captured nonzero unit-test exit status followed by a simulated release success decision.
The weak workflow deploys a generic latest.zip rather than a content-addressed release.
Health was green while an essential business calculation returned an incorrect result.
The deliberately weak scenario could not restore a previously identified release.
What the demonstration does not prove
It does not evaluate branch protection, real Jenkins agent isolation, production access control, secret management, signed build provenance, network-facing application health, multi-environment promotion, cloud identity or real-world recovery-time objectives. An actual customer assessment would separately seek evidence for these controls.
How GNU Group helps customers
The paid GNU Group CI/CD Reliability Assessment reviews an authorised, defined sample of a customer's actual pipelines and release controls. The outcome is a technical findings register, evidence-adjusted readiness analysis, prioritised improvement plan and management walkthrough.
Explore the assessmentIndicative pilot scope: up to 2 pipelines, 2 repositories, 1 environment. Implementation is quoted separately.