Everything written for this site is published under one byline, AI Agent Hub Editorial, because the work is shared and no single person should be credited for a claim three people checked. This page says what that byline stands for, what it covers, and what it refuses to do.
What this byline covers
Guides, category introductions, editor's notes, curated collections, glossary entries, and the moderation of reader reviews and questions. It does not cover the descriptions that arrive with a catalog listing; those are the publisher's material and the page says so.
How a claim gets in
Each guide starts from a job, not a product. If the honest answer is "you do not need a tool for this," the guide says that. Numbers on a page come from a named source: a vendor's own documentation, a public model card, a repository, or a leaderboard we name. When a number cannot be traced to a source, it does not go on the page. When a claim is our judgment rather than a fact, it is written as a judgment, and the guide says which part is which.
How this byline handles being wrong
Readers report errors through the contact form. A wrong license, a dead install prompt, or an outdated version is checked against the upstream repository and fixed. If a report is right and we disagree with the conclusion, we still fix the fact and keep the disagreement visible. The corrections policy describes the order in which reports are handled.
What this byline will not do
It will not rank a listing because someone paid for it. It will not turn a directory placement into a safety rating. It will not publish a reader review that is really an advertisement, and it will not publish personal data or unverifiable safety claims from a submission. It will not claim to have tested a model on hardware it never touched.
Voice
Plain sentences, concrete examples, and a bias toward saying when a tool is a bad fit. A guide that helps you reject ten tools and keep one is doing its job. Sentences that only sound impressive belong to a different kind of site.