Editorial standards

How a guide is written

Guides are drafted by an AI model against a fixed brief and checked by an automated review step against their cited sources before they publish. A human editor may not have read a guide before publication, and an automated check can miss an error. Report errors to info@custodytrack.io.

Each guide is written against a brief that fixes the subject, the question it must answer, and the ground it must not repeat from neighbouring guides. The writer is also given the verbatim text of the evidence rules the guides rely on most, and a set of settled points it may not contradict.

How claims are checked

An automated review step — a separate model pass, not a person — reads the draft against its sources and sorts every factual statement into one of three kinds. A checkable claim — a case holding, a rule or standard number, a statute, a statistic, a date, an attribution — must be supported by a cited source, or the sentence is rewritten or removed. A definitional statement needs no citation but must be accurate. An illustrative example must be labelled as one, so a worked scenario is never mistaken for a real matter.

Some checks are plain code rather than a model: a quotation of a rule must match the rule's actual text, and a list of overstatements (for example, calling a hash match proof) fails the guide outright. Automated checks reduce errors; they do not eliminate them, and an error a model believes can pass a model review. That is why reader reports matter.

Where the sources come from

Citations are restricted to an allowlist: courts and court records, government and law-enforcement agencies, standards bodies, and open-access scholarship. It currently holds 20 approved domains, and a citation to anything outside it fails the check — the count is read from the allowlist itself, so this page cannot overstate it.

Vendor marketing, content farms, and SEO blogs are not acceptable sources for anything a reader might rely on — including ours. Sources are listed in full at the foot of every guide, so you can check the reasoning rather than take our word for it.

What we will not say

A guide may not claim this product is certified or accredited by any standards body, may not promise any particular outcome in court, and may not overstate what the hash chain does. None of those would be true. Our records are tamper-evident — alteration is detectable, not prevented — and the design is informed by FRE 901 and NIST SP 800-86, which describes what we read rather than a conformance assessment anyone has performed. What a court makes of a record remains a question for the court.

These limits are enforced automatically. Every guide, and every page of this marketing site, is scanned before it can publish, and a prohibited claim fails the build rather than reaching a reader.

Corrections and updates

Guides are revised in place, keeping their original URL, and the revision date is published in each page's structured data. If you find an error — particularly a misstated holding, rule number, or standard — please tell us (or email info@custodytrack.io) and we will correct it.