KeyLens — API relay privacy checker

Find out if your AI API relay logs or leaks your keys and data

Problem

When using third-party AI API relays, users have no way to verify whether the provider logs requests or keeps sensitive data, yet they must expose accounts, VPS info, keys, and even real names and addresses to models. The community explicitly says "you cannot trust them not to log," but no actionable tool exists.

Solution

A local CLI/web tool that sends safe, identifiable probe requests containing no real sensitive data to test whether a relay echoes input verbatim, injects extra content, or caches responses. It outputs a risk score and alternatives, with a shareable report.

Users

Small teams and indie developers who regularly use API relays and shared AI proxies

MVP

Bundle a set of reusable privacy probe prompts; detect echo, injection, and caching of responses; generate a simple risk score; support a few common relay formats; never store user keys

Revenue

Free basic checks with paid detailed reports or batch-site subscription; self-hosted enterprise edition

Why now

AI API relays have exploded alongside AI adoption, privacy concerns are especially strong in Chinese-language communities, and no ready-made checker exists