# Editorial r1 rejection audit — evidence summary

Captured 2026-09-12 UTC. This record is descriptive evidence only; it does not establish a conversion outcome.

## Hosted-byte check

- Public URL: `http://15.204.83.60/CHL/PDP/editorial-r1/` responded `200` with `22,922` bytes.
- Public response SHA-256: `398711f96eacaadf584bbcc6758d42f646c1fcf1ea45b005618e4bb2f8be853c`.
- Candidate `projects/clean-health-lab/candidates/chl-editorial-r1-20260912/index.html` SHA-256: `398711f96eacaadf584bbcc6758d42f646c1fcf1ea45b005618e4bb2f8be853c`.
- Result: exact byte/hash match for the fetched HTML document.
- Hosted `styles.css` returned `200` / `24,674` bytes and matches candidate `styles.css` exactly: `b80f59b4e9be8c13731d5f67b2baf0f404b4161567b24e2efd7fda38c18965d1`.
- Hosted `app.js` returned `200` / `6,200` bytes and matches candidate `app.js` exactly: `b07e5413e6d19e2576400831811424fd85ad9a059b0cf9ce8f53a165232064df`.

## Browser captures and state counts

The `evidence/` directory contains exactly eight audit PNGs (not nine): desktop `1440x1000` and mobile `390x844`, each for top, proof initial, routines initial, and routines after the reveal action.

| Viewport | `#top` | `#proof` | `#routine` | `#routines` | UGC cards initial / reveal | `<video>` elements initial / reveal |
| --- | ---: | ---: | ---: | ---: | ---: | ---: |
| Desktop 1440x1000 | 102 | 3841 | 4521 | 5759 | 3 / 5 (of 5 total) | 1 / 1 |
| Mobile 390x844 | 114 | 5275 | 5995 | 7231 | 3 / 5 (of 5 total) | 1 / 1 |

Offsets are document pixels at capture time. The reveal control remains present after click; the two previously hidden UGC card buttons become visible.

### Font-ready confirmation and capture timing

The first capture run waited for `networkidle` but did not independently record `document.fonts` status; that is a timing limitation of its original metadata. A separate current verification on the same public bytes reported `document.fonts.status === "loaded"` before and after awaiting `document.fonts.ready` at both viewports. `document.fonts.check("500 48px Fraunces")` and `document.fonts.check("400 16px Manrope")` were both `true`; each loaded FontFace reported its expected family and weight range. Post-font-ready offsets exactly match the table above at both viewports. No recapture was necessary; this confirms current post-font-ready geometry, not the unrecorded first-run font event.

## Research-capture coverage

`fresh-design-20260912/evidence/competitor-research/screenshots/` contains 34 PNGs: 28 from seven four-view source sets plus six replacement views for Ritual and Dieux.

- Usable rendered visual evidence: seven references — Methyl-Life, Seeking Health, Needed, OSEA, Graza, Ritual, and Dieux.
- Captured but not usable for product-PDP visual conclusions: MethylPro (browser title: `Store unavailable`) and Seed (browser title: `Just a moment...` security-verification page).
- Text/indexed-only coverage in `RESEARCH.md`: AG1, IM8, Aesop, Jones Road, and Bellroy. These records support only the stated visible/indexed facts, not layout or visual-quality claims.
- `notes.md`, `sources.json`, and `RESEARCH.md` substantiate the above status split. Their own language says screenshots are momentary, blocked captures do not support visual conclusions, and reference patterns are hypotheses rather than conversion evidence.

## Incomplete baseline capture

`fresh-design-20260912/evidence/v61-baseline/mobile390-top.png` is a readable `390x844` PNG, but its large rounded hero-media frame renders as a blank pale field while surrounding logo, offer bar, product text, review line, gallery thumbnails, and body copy are visible. It is therefore incomplete for a hero-image/visual comparison. This image can evidence that the captured frame was blank; it cannot establish that blank media was an intended v61 design state or a baseline design defect.

## Isolated detector artifact

The candidate detector output is saved at `evidence/impeccable-detector-assessment-b.json`. It contains 12 findings: 3 warnings and 9 advisories (five rule categories).