Safety methodology
Food safety is a separate layer with separate authorities, and it outranks every cooking recommendation on this site.
Safety is a separate layer with separate authorities
Food-safety rules on this site come from national food-safety authorities. They are never derived from a cooking observation, never from a recipe, and never from a manufacturer's chart — a manufacturer is authoritative about its own appliance and is not a food-safety authority.
This is enforced in the code rather than promised. A validation rule checks the source type behind every published safety rule, and a microbiological claim supported only by culinary testing fails the build rather than shipping with a caveat.
Safety outranks cooking advice, and suppression is total
Where a cooking recommendation cannot be reconciled with a rule that governs it, the recommendation is not shown. Not shown with a warning beside it, not greyed out, not at all.
A time that cannot produce a safe result is not made safe by a note underneath it, and showing both invites a reader to follow the number and skim the note. There is exactly one code path for this decision and it suppresses.
Absence of evidence is never converted into safety
Where we have not established whether something is safe, the answer is 'not established' — a terminal state that renders as 'we do not know', never as a cautious yes and never as a silent omission.
This is architecture rather than policy. If 'no record exists' and 'checked and safe' were the same value, every gap in the catalogue would read as an all-clear and the product would get more dangerous as it grew. They are different values, and a runtime check throws if code ever produces a permissive answer without a record behind it.
We publish disagreement rather than resolving it silently
UK and US authorities state different poultry temperatures. They are not in conflict about the microbiology — the figures sit on the same lethality curve at different points — and a reader who has met both numbers deserves to be told why they differ rather than being handed one of them.
Where sources disagree, the record carries both positions attributed, the likely reason, and which position AirfryerHQ follows. Where we do not adjudicate, we say that too.
Where we stop
AirfryerHQ gives no repair advice for electrical faults, and there is no field in the data model that could hold one. An appliance drawing between 1.4 and 2.7 kilowatts through a heating element is not something to investigate with a screwdriver.
Troubleshooting pages for electrical and smoke symptoms carry a required escalation: stop, switch off at the socket, unplug, do not use it again, contact the manufacturer. That requirement is enforced by the schema, not left to an author's judgement on the day.