Agent Ops / Standard term
Consent screen
A consent screen is the page an account provider shows the first time you grant software access, listing the app name and the exact permissions being requested so you can approve or deny before any access begins.
A consent screen is the page an account provider shows the first time you grant software access, listing the app name and the exact permissions being requested so you can approve or deny before any access begins. It is the moment where you decide what an outside app may reach inside your account. Say you connect an AI assistant to your email so it can draft replies. The provider stops and shows a screen naming the assistant and the scopes it asked for, such as reading your messages and creating drafts. You read that list, confirm it matches what you intend, and click approve; only then does the provider hand the assistant a token to act within those limits. Open Authorization (OAuth) is the protocol behind this handoff.
Builder example
When you wire an AI agent into a calendar, inbox, or file store, the consent screen is the last place a person verifies what the agent can touch before access is live. If you request broad scopes for convenience, the screen warns the user and they may decline, or worse, approve more reach than the task needs. Tell your assistant to request only the scopes one task requires, then read the consent screen aloud and confirm each line matches the job before you approve.
Common confusion: The consent screen shows what an app is asking for; it does not narrow the request on its own. What you approve is exactly what the app listed, so an overly broad ask grants overly broad access. Read each scope before clicking, because approval is the point where the access begins.