91.7%
↓ 3.1% from prior window
No OAuth, private access, or stored credentials.
Evidence-first triage across builds, tests, and deployments.
main· 4d8f2c1
Apply a local policy to the selected run. The diagnosis and scorecard stay fixed; only the final release action can change.
Browser-only simulation · no release state changesProtected-branch regression evidence satisfies the configured release policy.
Diagnosis remains Probable code regression · 91% confidence| Run | Change | Status | Duration | Finished |
|---|---|---|---|---|
| Harden payment retry policy | failed | 7m 42s | 4 min ago | |
| Cache catalog lookups | passed | 5m 18s | 38 min ago | |
| Add regional tax mapping | failed | 9m 04s | 1h ago | |
| Update checkout telemetry | failed | 5m 06s | 2h ago | |
| Bump validation runtime | cancelled | 1m 12s | 3h ago |
Public workflow, job, commit, queue, and branch signals enter a deterministic scoring engine. Optional AI can summarize the result later; it cannot change a release decision.
Review the implementationOutcome, queue, workflow history, and changed paths; retry outcome when verified.
Regression, flaky test, runner pressure, or provider incident.
Protected branches block on reproducible code regressions.
verdict code-regression
confidence 91
release BLOCK
engine deterministic/v1.1