Editorial Standards.

Sourcing.

Every Wolverine article cites at least three primary sources. Primary sources are: peer-reviewed studies (PubMed, PubMed Central, preprint servers with clear peer-review status noted), regulatory filings (FDA, Federal Register, FTC enforcement actions), and official registries (ClinicalTrials.gov, WADA Prohibited List, PCAC hearing transcripts).

Secondary sources — editorials, newsletters, competitor commentary — may appear as context but never as the sole support for a factual claim. We trace every secondary source claim to its primary before publishing. Supplement-vendor blogs are not acceptable sources.

The ingestion pipeline covers four signal buckets: regulatory and scientific primary sources, audience platforms (podcasts, community discussion), industry signal, and editorial peers. Each source type has its own ingestion policy in infra/sources.md.

Attribution.

Every substantive claim links to its source via inline <a href> or appears in the article's structured Sources block at the foot of the page. The Sources block is the canonical reference list; inline links are navigation aids.

We apply strict cite-don't-republish discipline. Cited excerpts are ≤50 words — the fair-use ceiling enforced at the schema level. We store the source URL, access timestamp, content hash, and license basis (public-domain, fair-use, licensed, or rss-permitted) for every cited source.

We do not republish third-party content wholesale. We do not reproduce full article bodies, full study text, or full podcast transcripts. We summarise, quote selectively, and link.

Hedging.

Claim language is evidence-graded and context-appropriate. We apply consistent hedging that distinguishes the strength of available evidence:

• "Human evidence supports" — for claims backed by controlled human trials.
• "Studies suggest" — for observational human data or mixed-quality trials.
• "Animal models show" — for preclinical data not yet replicated in humans.
• "Preliminary evidence indicates" — for single studies, small samples, or early-phase data.

Forbidden claim verbs: cures, heals, boosts, transforms, reverses. These terms make efficacy claims that exceed available evidence and cross the science-journalism lane into clinical recommendation territory. Use mechanism-of-action language and evidence grading instead.

Lane discipline.

Wolverine content is science journalism, not medical advice. We explain mechanisms, analyse regulatory developments, and report on industry and research. We do not recommend, prescribe, or counsel on personal health decisions.

Every article carries the Disclaimer component confirming this posture. The boundary is enforced by editorial review before publish and by automated forbidden-phrase linting in the build pipeline. If an article contains language that crosses into clinical recommendation, it does not ship.

Corrections.

We do not delete history. Corrections are published with a visible <aside> notice at the head of the affected article noting what changed, what it was changed to, and when. Substantial updates — new regulatory information, retracted source — trigger a revised lastReviewedAt date and a "What changed" footer block.

All historical versions are preserved in version control. The public article always shows the latest reviewed version. Corrections are logged.

To report a factual error: corrections@wolverine.health.

DMCA and takedown.

If you believe Wolverine content infringes your copyright, send a takedown notice to dmca@wolverine.health. We follow the full DMCA safe-harbour procedure. See the Content Disclaimer for the full procedure and contact details.