Articles from the App Use team on transparent agent–app control — the ideas behind the protocol, why audit and human consent matter, and practical guides for making your own app agent-ready. Subscribe via Atom.
Today's agents either screen-scrape (brittle, blind to the OS) or shell out to opaque CLIs. There's a third path: let the app declare what it can do and have the agent drive it through one open protocol — transparently.
When an agent acts inside your software, "trust me" isn't an answer. App Use makes every action observable, records it in a tamper-evident chain, and stops dangerous steps for a human to approve.
A practical walk-through: declare your screens and elements, implement one interface, host it, and watch an agent describe and drive your app — with the SDK handling transport, consent, and audit.
New articles are added over time. Want to contribute a real-world report instead? See Field Notes.
If there's an aspect of agent–app control, the protocol, or adoption you'd like written up, tell us.