5 posts
"May it ship?" should not be answered in a chat. The stlc board on orders checkout, and a QA release gate: a report verdict derived from its gate rows, rejected by lint if it disagrees, which the dev deploy step reads before checkout ships.
The cheapest fix for a red AI test is to hide it. On checkout a red test is labelled, not fixed: four triage labels, a heal guard that diffs every fix, quarantine that expires in 30 days, and a defect that closes only on a newer passing run.
You cannot tell whether AI-written tests cover the spec. The first stlc run on orders checkout: requirement ids, a traceability table created with zero rows where each column has one writing step, a case review, and ten scripts.
Forty green AI-written tests shipped a bug and nobody could say what they proved. Five questions to ask any AI test pipeline, and the route stlc takes from docs to a verdict, where a feature's state is computed from the files each step wrote.
sdlc and stlc meet at one file out and one report back: G9 writes qa-handoff.md with the REQ and AC ids, stlc returns a QA report whose overall verdict is computed from its gate lines, and G10 refuses to deploy without a passing report dated after the shipped commit's G8 run.
