Scope Disclosure

Scope disclosure is an AI UX pattern that states in plain language which accounts, files, tools, and actions an agent can reach, at connect time, on demand, and when scope changes. It turns permissions into a readable contract instead of a buried OAuth dump.

Share

Interactive demo

Settings → Agents

Inbox assistant

Not connected yet. Link a mailbox to get started.

Gmail

Work inbox · you@company.com

Overview

The design problem

How might we design scope disclosure so people can trust and act on AI output?

Use this pattern

When this pattern fits

  • Essential for AI agents, OAuth-based assistants, and any system that acts on behalf of users, where plain-language scope disclosure builds trust and reduces post-hoc security risk.

Avoid this pattern

When to skip or lighten it

  • Local toys with no external permissions to disclose.
  • When legal requires full policy text but you never provide a short summary (always give both).
  • One-time scripts with no retained agent identity.

States

State model coming soon

Key UX elements

Key UX elements coming soon

Anti-patterns to avoid

  • Allow-all modals with no capability list.

  • Disclosing scope once at install and never again after expansion.

  • Jargon-only scopes (“access user.info.email.write”) with no human translation.

  • No place in settings to re-read current scope.

How products use it

ProductImplementation
ChatGPT connectorsPermission summaries when linking apps and mail.
Claude connectors / computer useCapability boundaries explained around tools and browsing.
GitHub Apps / OAuthScope lists at install with account-level review.
Enterprise AI gatewaysAdmin-visible agent capability contracts per team.

Implementation

Copy this prompt to generate a production-ready implementation in Cursor, Claude Code, Lovable, or any AI coding agent.

Generate a production-ready implementation of the "Scope Disclosure" AI interface design pattern.

Pattern Definition:

Frequently asked questions

What is scope disclosure for AI agents?

Scope disclosure is a plain-language summary of what an agent can access and do, shown when connecting, available later in settings, and repeated when permissions grow.

When should scope be shown again?

At first grant, whenever scope expands, and on a review cadence for long-lived agents. Do not assume users remember a modal from months ago.

How does scope disclosure differ from granular consent?

Disclosure explains the contract. Granular consent lets users change it capability by capability. Ship explanation and controls together.

Weekly AI UX in your inbox

Weekly AI interface UX notes and resources on Substack, no spam, unsubscribe anytime.

Subscribe on Substack