The work
Three product pages pointed visitors to a generic educational guide. We added practical first-evaluation guides for Harness, Command Center and Odin Next, plus a setup checklist. The scope covered website content, links and presentation. It did not create product accounts, change a customer environment or establish new commercial terms.
A defect the checks initially missed
The first staged candidate passed its automated page checks. Visual review then found that the guide’s sticky setup panel could slide behind the global navigation when scrolled on desktop. We held that candidate and made a forward correction before promotion.
| Position | Before | After |
|---|---|---|
| Header bottom | 135 px | 135 px |
| Setup panel top | 40 px | 160 px |
| Panel behind header? | Yes | No |
The follow-up check explicitly scrolled the page and compared the two positions. The original layout fails that check; the corrected layout passes. A passing automated sweep was useful evidence, but it had not covered this state until we added the check.
What we checked on the release
For source commit 3f9bac91ed37a2f8c526cca8a44acb3c9c75d455, the source test run passed 197 tests and the build produced 33 pages. After staging and promotion, public verification matched 76 static files to their build bytes, checked 63 redirects and four API refusal cases, and completed 48 browser cases across Chrome, Firefox and WebKit.
The new pages were checked with JavaScript enabled and disabled, at 320, 390 and 1440 pixels, including links, keyboard access and the scrolled panel. The public observations were recorded on 19 October 2026 between 20:49:02 and 20:49:45 UTC.
Inspect the public evidence summary. It names the source, commands, measurement window, result and limits. This is a summary published by Odin from retained receipts; it does not make the private repository or full test harness public.
What this does not prove
This is our own website delivery, not a customer case study or an independent benchmark. It does not demonstrate customer productivity, sales conversion, a product backend or a customer deployment. The source and measurement window matter: the website can change after this release.