reportlisted

# report — the standardized run manifest Follow the shared contract in `../README.md`. This skill produces the **one envelope every QAIA plugin shares**: `.qaia/reports/<US-ID>/manifest.json`, defined by `../../OUTPUT-CONTRACT.md` (that path is in the QAIA source repository, not in the installed plugin — the fields it defines are restated wherever this skill needs them). It never invents data — it *projects* the existing journey artifacts into the common schema so `qaia-score`, an export, or CI can read any run the same way. ## Prerequisite A generated test book in `.qaia/testbooks/<US-ID>/` (with its `synthesis.md`, `coverage-matrix.md`, and the `03-design.md` / `04-priorities.md` checkpoints). If it is absent, say which step is missing and offer `testbook-generate` — never emit a manifest with guessed counts. ## Steps 1. **Read the source artifacts**, never re-generate them: - the `.feature` files (count scenario blocks, priority/`@negative`/`@smoke`/Outline tags, technique tags, `@oracle:*` provenance); **`total` counts executable cases, not blocks.** A `Scenario Outline` with N `Examples` rows counts N — that is what a runner will execute, and what `testbook-export` already projects as N rows. `outlines` carries the number of un-expanded blocks. Counting blocks instead made a book of 10 Outlines × 6 examples read as 10 scenarios in the manifest and 60 rows in the reviewer's spreadsheet, with every ratio — negative, low-confidence, `byPriority`