Historical text report · Reviewed October 6, 2026
The records behind the makeup audit
The preserved report describes 155 SKU entries in 81 approximate groups across four brands. It is a record of an earlier automated text screen, with incomplete saved ingredient passages.
| Brand | SKU entries | Approximate groups |
|---|---|---|
| Kylie Cosmetics | 20 | 20 |
| Huda Beauty | 13 | 12 |
| rhode | 63 | 35 |
| Fenty Beauty | 59 | 14 |
What survives, and what is missing
The internal report retains grouped SKU identifiers, counters, legacy keyword annotations and ingredient-text excerpts capped at 400 characters. The original raw audit_results.json, original page HTML and capture timestamps are absent. The earlier articles assigned a July 30 collection date; the surviving report cannot independently verify that exact date.
The downloadable archive description preserves every group’s identifiers, stored character count and text fingerprint, plus the report’s SHA-256 fingerprint. It omits full manufacturer passages and the withdrawn particle annotations. Every group’s complete-list status is unverified. This is not a newly collected or corrected ingredient-match dataset.
Why the old prevalence cannot be repaired from this file
A shortened excerpt can omit an ingredient, end halfway through a name or contain a mixture of kit components. The old grouping also approximated formula identity across shades. Running a better matcher on the first 400 characters cannot reconstruct the missing ending or establish that the grouping was correct.
A current product page can verify a selected product observation, but it cannot retroactively fill the historical raw input. The new brand guides therefore cite selected manufacturer listings separately. They do not rename all archived groups as currently checked formulas.
Reproducing this description
The repository generator scripts/beauty-audit/review-makeup.mjs reads the preserved report without modifying it, checks grouped counters, and writes this page’s coverage module and JSON description. It performs no new crawl, name screening, particle assay or compliance determination. The method version is makeup-archive-description-v1.
A fuller future study would need complete source lists, capture times, market and variant identification, separate kit components and an explicit formula-comparison method. A particle claim additionally needs evidence of physical form, size and relevant measurement conditions. The metadata download cannot supply those observations.
Read the corrected makeup comparison for selected products and a practical label-checking workflow.