Inspect reports and test results#

Two useful artifacts demonstrate the reporting boundary: an illustrative portable bug-report bundle, and a reviewed test-run report generated from an actual run.

Generate the portable example#

After workspace setup:

pnpm example:report

Open dist/report-example/index.html and inspect report.json beside it. Preserve the whole directory when copying a bundle that includes attachments. The documentation artifact exposes this generated example with the same illustrative status.

The example source defines a passing checkout attempt with an incidental menu problem. It adds a problem comment, promotes that comment to a bug report, and redacts a diagnostic session value before export. The promoted bug does not inherit checkout's expectation or a verified reproduction claim.

The data is authored: no browser session is recorded, no screenshot is collected, and no tracker ticket is created. The report APIs and sanitization path are real. Portable report guidance describes the exact processing contract.

In the built documentation artifact, open the generated illustrative report.

Explore both sample reports#

The sample report gallery links the portable bug report and an interactive sample test-run report. Both are included in the static website artifact, including the viewer resources and structured JSON. Use the sample navigation to return to the presentation site.

The test-run sample contains authored passing, failing, and retried profile checks, with diagnostic observations and recorded interactions. Select results, compare attempts, filter outcomes, and create a local bug-report export. These are illustrative records, not evidence of a live execution. The sample generator uses the maintained report renderer.

Review a real run#

The first test-run workflow produces a checkpoint and, after policy review, an HTML run report. Use its title/tag/component/level search, outcome filters, hierarchy and attempt history. Compare intended steps with actual interactions; inspect hook diagnostics separately from behavioral evidence.

Required missing proof makes an attempt incomplete even when a behavior failure was observed. An earlier failed attempt remains selectable after a successful retry. Create bug opens an editable local draft, including an observed failure with incomplete proof; it does not submit automatically.

Before sharing, sanitize narrative and attachment bytes. Raw run output is private execution state, not a reviewed export. See serialized formats for checkpoint versus bundle distinctions.