Agentic control
Checkpoints
Use when a long agent session changes a lot and people want a safe point to roll back to.
Named restore points. Confirm inline and roll back with re-verification.
Part of Agentic control. The stage where the agent stops answering and starts doing. It covers the whole handover: agreeing what the agent may do, watching it work, stepping in, and checking what changed afterwards. Reach for these as soon as an action touches a customer, money or a record.
Do
- Name checkpoints after what happened, such as After Acme replies, with a time and a measure of state.
- Confirm inline, saying how many actions will be rolled back.
- Re-check the live state before restoring, and say that the agent is doing it.
Don't
- Don't open a modal to confirm. The strip under the row is enough.
- Don't offer Restore on the current checkpoint.
- Don't restore silently. Report how many actions were reversed.
States
- Idle: A timeline of named checkpoints. The current one has a pulsing accent dot.
- Restore offered: Hovering an earlier checkpoint shows a Restore button on that row.
- Confirming: A neutral strip opens under the row with the number of actions to roll back, and Restore or Keep going.
- Verifying: Later checkpoints collapse and a thinking indicator says the agent is verifying ticket states.
- Restored: The chosen checkpoint becomes current, with a caption saying how many actions were reversed.
Usage
import { CheckpointPattern } from "./halaska-kit";
<CheckpointPattern />
Built from
Agent Glyph, Button, Stack, Thinking Indicator, Typography
Related patterns
Halaska UI · Components · Patterns · Screens · Install