UtilForge.org

The 'What Not to Paste into AI' check is my favourite of the bunch — most people have no idea how much personal detail sits in a block of text until something flags it. Running everything on-device is the only way a tool like this stays trustworthy. One question for the maker: do the checks work fully offline, or do some need to fetch fresh scam-pattern data periodically?

Running the scam checks on the device itself is the right call for privacy tools, people are rightly nervous about pasting suspicious links into a random website. Curious how you handle lookups that need the network, like checking where a shady link actually redirects. Great launch!

UtilForge began with one question at a kitchen table: how does an older relative check a message that asks for money, without sending that message to anyone? Every UtilForge check runs on the person's own phone or computer, so what they paste stays there. Forty-two tools later, the same rule holds for artificial intelligence: Hide Names from AI swaps names and numbers for labels before you paste into any assistant, and Prompt Injection Check finds hidden instructions in a page before your assistant reads them. The tools can be used without paying, and the large-print book may be printed and shared by any library or community group under its Creative Commons licence. Questions and suggestions are welcome; every message is read by a person.