About this skill

Problem

Many repositories are not missing tests; they are missing merge gates that cannot be skipped. Lint failures get commented out, tests get skipped, production credentials end up in config files, and deployments have no rollback path. This is especially visible with AI agents, which can generate code but should not replace the team’s rule that every change must be verifiable.

How This Skill Works

It treats CI/CD as a quality-gate system rather than a pile of ad hoc scripts:
- Pre-merge checks: lint, type checking, unit tests, builds, and dependency audits must pass, and failures block merge.
- Shift left: static analysis, tests, and deployment verification are ordered by risk so bugs and misconfiguration are caught earlier.
- Deployment safety: each PR gets a preview environment, risky features ship behind feature flags, and releases keep a rollback path.
- Feedback loop: CI failure details are routed back to agents or engineers for fixing, retesting, and improving coverage instead of being ignored.
- Engineering governance: branch protection, required status checks, secret separation, and a Build Cop pattern prevent broken builds from piling up.

Boundaries

It fits new project setup, completing automation checks, debugging CI failures, and configuring deployment pipelines. The source does not provide complete platform templates, so real adoption still needs GitHub Actions, database, E2E, and secrets-manager details. Treat it as a set of rules, sequencing, and enforcement expectations, not as a test framework or release platform by itself.

Screenshots

Use Cases

When bootstrapping a new project, configure PR gates for lint, types, tests, and builds so failures block merge.
When tests keep failing, trace flaky tests or failed steps, fix the tests instead of skipping them, and restore branch protection.
Before shipping risky features, set up feature flags, preview deployments, and rollback switches instead of full rollout by default.
When CI runs over ten minutes, optimize the pipeline with caching and parallelism instead of deleting checks or weakening gates.

Best For

Backend engineers initializing a new repo who need quality gates and branch protection from day one.
DevOps engineers maintaining CI who need to trace failures and use CI feedback to drive fixes.
Engineers working with AI coding agents who need agent output covered by tests, builds, and rollback checks.
Product engineers owning release workflows who need feature flags, preview deployments, and rollback plans.