Research and sourcing
We prefer primary sources such as official documentation, policy pages, standards bodies, and direct product materials. Material claims are checked at the substantive review date shown on the page. External sources can change afterward.
AI-assisted workflow
Automation may collect reproducible first-party measurements, organize source notes, draft outlines, check links, and flag stale passages. It is not allowed to invent tests, quotes, prices, usage, results, authorship, or customer experiences. Model-written quality scores are not editorial proof. During recovery, automated prose publishing is disabled; an indexable report requires a public evidence bundle, stated limits, a dated release record, and a documented remediation decision.
Review labels
We distinguish hands-on reviews from selection guides. A product is not described as tested unless a documented evaluation took place. Vendor claims, observed behavior, and editorial judgments should be recognizable as different kinds of evidence.
Commercial independence
Affiliate availability does not determine whether a product is included or scored. Sponsored material, if introduced, will be clearly labeled and separated from independent editorial conclusions.
Advertising eligibility
Advertising remains disabled until platform approval. If enabled later, ads may appear only beside substantive publisher reports. Tools, indexes, policies, errors, health checks, APIs, and archive pages are not advertising surfaces.
Corrections and updates
Material errors are corrected promptly. Substantive updates change the review date; typo fixes do not. Readers and vendors can report a concern through the contact page. We evaluate corrections against evidence rather than commercial pressure.