judgelisted

# judge — the gap between a promise and what claims to keep it Follow the shared contract in `../../README.md`. QAIA is not a test generator. It is an engine for the gap between **a promise** and **what claims to keep it**. `judge` is the face where the second term already exists and someone wants to know whether it holds up. **It judges what other people wrote, first-class.** Not as an afterthought, not in a special mode. The tool once applied this project's own conventions by default and hid the exception behind a flag; pointed at 257 Gherkin books written elsewhere it returned **0 PASS**, and 493 of its 666 findings were about conventions that do not exist in Gherkin. The default is now the universal scale, and the same corpus returns **102 PASS, median 77, 150 findings**. ## What you can hand it | You have | It answers | |---|---| | `.feature` files, any origin | are these tests, or questions dressed as tests? | | a Playwright suite | do its assertions carry weight, and are they traceable to anything? | | an OpenAPI spec **and** a suite | do they still agree, or has one drifted? | | any of the above **plus** the source requirement | plus: coverage, business correctness, declared ambiguity | Without the source requirement, the requirement-dependent dimensions are marked **not assessable** — never guessed, never scored zero. ## The order, and it is not negotiable **1 — Run the pinned scorer. Do not re-implement it.** ```bash python "${CLAUDE_PLUGIN_ROOT}/scripts/st