# &lt;Chapter title&gt; · verification receipt · &lt;date&gt;

> Copy this directory to `verification/<chapter-id>-<yyyy-mm-dd>/`, replace every
> angle-bracketed value and every sentence below, and delete this quote. A
> receipt records what was actually run against the delivered tree — not what
> the plan said would be run. If something was not checked, it belongs under
> *What was not checked*, never omitted.

Scope: &lt;what this chapter changed, in one or two sentences&gt;.
[state.json](state.json) holds the exact counts, hashes and tool results
summarised here.

## What was checked

- **Build freshness.** `build-eve-oshun.py --check` on the delivered tree, with
  the slide, source and guide counts it reported.
- **Narration.** Which slides were rendered and why, on which device, and the
  exact `--check --ids` result. Name the tracks left untouched.
- **PDFs.** Which guides were exported, with the page count and SHA-256 of each,
  and the checks the export itself asserts.
- **Browser rendering.** `eve-edition-review.mjs --guide <id>`: the viewports
  swept, the findings, and where the report and contact sheet are. Name any
  slide that needed a change to fit, and what changed.
- **Captures.** For every new capture: the component path, the pinned revision,
  the fixture, and the browser that rendered it. A capture harness is bound to
  its browser build — a rerun on a different Chromium produces different bytes
  and breaks the slide pins.
- **Suites and lint.** Python and Node counts, every remaining failure with
  whether it is present on the unchanged baseline, and Ruff, ESLint and Prettier
  on the changed files.

## What was not checked

- &lt;Each thing a reader might reasonably assume was checked and was not.&gt;
