Active incidents
Do not deploy into a live incident without explicit context.
Deploy Gate runs in GitHub Actions and posts a PR verdict using the production signals your reviewer cannot see in the diff.
No blocking issues. Safe to deploy.
Each signal is simple enough to explain in a code review. The value is putting the signal bundle next to the merge decision.
Do not deploy into a live incident without explicit context.
Recent failures and churn make the next deploy riskier.
Health context helps avoid pushing into a fragile service.
Latency and error signals become part of the PR discussion.
Lockfile movement is reviewed at the deploy boundary.
No blocking issues. Safe to deploy.
Start with GitHub Actions. Add PagerDuty and Datadog when you want live incident, SLO, and health context.
The gate evaluates the signals Strake can currently see and reports missing sources instead of pretending coverage is complete.
The PR gets a GO / HOLD / CRITICAL verdict plus the signal breakdown, so the team knows why the gate made the call.
Strake does not need a black box to be useful. The gate should be boringly clear: any critical signal is critical, warnings hold the deploy for review, and clean evaluated signals produce GO.
Install the GitHub Action, open a pull request, and judge the verdict before adopting anything else.