Cyber Decision Lab is the free, browser-local companion to Paragamix.Cyber. While building its message and link checks, one design question kept returning: users understandably want a simple safe or scam answer, but metadata and language patterns rarely justify that level of certainty.

The current approach is to show the observable evidence, explain what the check cannot determine, and suggest a next action. It does not present itself as malware scanning, sender authentication or a substitute for incident response. Inputs remain in the browser.

The tradeoff is usability. A careful answer can feel less satisfying than a large risk score, even when it is more honest.

For people building security or trust features: what would you trust more in a free checker a risk score, a short evidence list, or an explicit refusal to classify an uncertain message? Have you seen a product communicate this uncertainty particularly well?