Consent Screen
User-facing interface where a human explicitly authorizes an AI agent to access specific tools within defined boundaries. Time-boxed, revocable, and access-aware.
A consent screen is the user-facing interface where a human explicitly authorizes an AI agent to access specific tools within defined boundaries. It is time-boxed (expires after a configurable period), revocable at any time, and access-aware — the tools and trust tier shown are capped by the user's RBAC.
What consent screens show
- Available tools: which MCP servers and tools the agent will access
- Trust tier: level of automation being granted
- Duration: how long the consent remains valid
- RBAC caps: what the user is actually allowed to do (the ceiling)
Consent vs HITL
Consent is a one-time upfront authorization. Human-in-the-Loop is a per-operation gate for sensitive calls. Both work together: consent defines the agent's operating envelope; HITL gates individual operations that exceed normal risk thresholds.
Related terms
Authorization that evaluates permissions at the individual resource or action level, rather than broadly by user type or role.
Security pattern requiring a human reviewer to approve sensitive operations before an agent executes them. Single-use approval consumed on the next matching call.
Graduated levels of trust assigned to AI agents: low trust requires explicit consent for every tool, high trust auto-approves common patterns. Capped by the user's RBAC permissions.