WhoUsedMyKey

WhoUsedMyKey

Approve what your AI agents do

0 ratings
Free

Rating summary

Details

Released
November 7, 2026
Updated
November 9, 2026

Features

WhoUsedMyKey screenshot #1 for iPhone
WhoUsedMyKey screenshot #2 for iPhone
WhoUsedMyKey screenshot #3 for iPhone
WhoUsedMyKey screenshot #4 for iPhone
WhoUsedMyKey screenshot #5 for iPhone
iphone
ipad
🖼️Get Icon
Icons↘︎

About

AI coding agents like Claude Code and Cursor run commands on your laptop, and they can reach your API keys, cloud logins and .env files. WhoUsedMyKey shows you what they touch, and lets you decide on the risky moves from your phone. APPROVE RISKY ACTIONS When an agent tries something risky, such as a force-push to main, deleting files or a deploy, it pauses and your phone asks. Approve or deny in a tap, with Face ID or your passcode for high-risk actions. If nobody answers in time, the action is denied. SEE EVERY KEY YOUR AGENTS CAN REACH See which secrets each agent on each laptop can read, when each one was last used, and by which command. Only names and locations are reported. Key values never leave your laptop. SET A RULE FOR EACH KEY Allow a key, make agents ask you first, or block it outright. Loosening a rule needs your phone’s signature, and the laptop checks that signature itself. ANSWER YOUR AGENTS When an agent has a question or needs permission, answer it from your phone, or send it back to the terminal. A FULL ACTIVITY FEED Search and filter what your agents did, what was allowed and what was stopped. SIGNED ON YOUR PHONE Every decision is signed by a key that never leaves your phone, so your laptop can tell a decision really came from you, not from our server. HOW IT WORKS Install the free WhoUsedMyKey command-line tool on your laptop, run wumk login, and approve the code it shows in the app. No analytics, no ads and no tracking. You can delete your account at any time in Settings.
Show more

What's New in WhoUsedMyKey

1.0.1

November 9, 2026

Approvals are now much easier to read. • See what a command does at a glance: approvals now start with the AI agent's own one-line description, clearly marked as unverified. • Long commands are folded, with their full size shown, so you can expand and read every line before deciding. • The command no longer appears twice on the approval screen. • Permission prompts show the same description and folded view. Descriptions appear when your AI agent provides one, and need the latest wumk CLI.

More