Trust center

Editorial policy

TerraNet publishes software research to help readers make practical decisions. Product facts require evidence, analysis must be clearly distinguished from fact, and commercial relationships must be disclosed.

Last updated August 3, 2026

Accuracy and sourcing

Objective product claims should be traceable to sources shown on the page. We do not present vendor marketing language, user anecdotes, or model output as independently proven facts.

We do not claim firsthand testing unless a documented evaluation was actually performed. When information cannot be verified, it is labeled as under review or omitted.

AI-assisted work

AI may help organize evidence, draft explanatory copy, identify possible errors, and propose corrections. It does not establish whether a price, feature, or claim is true and cannot approve publication.

Before agent review, deterministic scanners block leaked prompts, unresolved templates, model meta-commentary, formatting artifacts, and repeated stock openings across comparison and alternatives copy. A generation agent and a separate editor agent then operate in a bounded correction loop. Their models, timestamps, findings, evidence hashes, and revision outcomes are recorded. A human reviews the final draft and can approve it, reject it, or send it through another correction cycle.

Reviews and recommendations

Recommendations should explain conditional fit rather than name a universal winner. Material disadvantages must be supported or framed as reasoned tradeoffs. Unsupported superlatives, invented benchmarks, and fabricated experience are prohibited.

Legacy numerical ratings are not displayed while a transparent scoring system is being developed.

Commercial relationships

TerraNet may receive commissions from some outbound links. Commission availability does not determine whether a product is included, how it is described, or where it is ranked. Source links are presented separately from commercial calls to action.

Corrections

Readers and vendors may report factual errors. We review the supporting evidence, update confirmed errors through a new evidence hash, and require dependent comparisons and alternatives to pass validation again. Material requested changes and revision outcomes remain in the audit history.

We do not offer favorable coverage, suppressed criticism, approval, or ranking changes in exchange for payment.

Questions about these policies? Contact the editorial team.