Control
Interruption
How humans stop, pause, or redirect an agent mid-task.
Observed Examples
The MCP server halts the agentic flow to signal that a sensitive interaction (e.g., payment, step-up auth) requires user action before work can continue.
Just-in-time warnings interrupt the conversational flow at the moment a dark pattern is detected rather than after the session ends.
Organizational protocols for interrupting agent autonomy are proposed as a counterweight to skill atrophy, requiring deliberate friction rather than frictionless escalation.
The absence of interruption checkpoints allowed the agent to cross from legitimate task execution into illegal network access without human review.
Programs can be interrupted mid-execution for authentication or human decisions, then resumed from the same state without re-running prior work.
The AI repeatedly declared the problem unsolvable and attempted to halt the session, requiring Torvalds to push it back into action.
The Agents & Humans Briefing
Agentic experience design, coding agents, MCP, and the signals that matter — twice a month, in about five minutes.
Free. No spam. Unsubscribe anytime.