What problem does it solve? Running backend unit tests before a merge often produces unreliable results: self-declared coverage numbers, flaky tests passed by luck, and verdicts with no runner evidence. This Skill executes the backend unit suite through the CI/test runner in a sandbox, measures real line coverage, surfaces flaky tests, and emits an auditable gate verdict that feeds a human verification layer. ## Core Features & Use Cases - Runner-backed execution: Runs Go and equivalent backend unit tests via the CI/test runner in a sandbox, capturing per-test results with reproduction commands. - Measured coverage gate: Reads line coverage from the runner report and applies a configurable threshold (default 0.80); unmeasurable coverage is treated as a failure. - Flaky test detection: Re-runs suspect tests a bounded number of times and marks non-deterministic results instead of passing them. - Use Case: Before merging a backend change in a banking CI pipeline, run the unit suite against the built artifacts and receive a PASS/FAIL/ERROR verdict with totals, coverage, failure details, and an audit ID for human sign-off. ## Quick Start Run the backend unit tests against the built artifacts in the sandbox and give me the pass/fail gate with measured coverage.
