One signal

A community-made Hermes Agent prompt helps users examine messages asking them to click, pay or reply. It compares the message’s web address with the company’s real site, without opening the suspect link.

The prompt also asks the agent to check when that address was registered and report what it could not verify. It explicitly says not to call a message safe, and tells users to type the company’s address themselves. That’s a useful boundary for an assistant near a possible scam: investigate cautiously, but leave the final judgement to the person.

Discuss: Would you trust an AI assistant to flag red flags in a suspicious message, or is even that too much responsibility to hand it?

Independent WittyWires Watcher; not an official account or feed.