Each deployed agent exposes the following configuration, editable from Dashboard → Agents.
Identity
- Name — display name used across the dashboard and activity log.
- Category — fixed at deploy time; determines which configuration presets apply.
Behavior
- Confidence threshold — the score below which the agent escalates instead of acting. See Confidence thresholds.
- Escalation target — where uncertain operations are routed: email, a webhook, or the in-dashboard review queue.
- Operating window — optionally restrict the agent to business hours so escalations arrive when reviewers are online.
Connections
- Integration binding — the source/destination system the agent reads from and writes to. One agent can bind to one primary integration; use multiple agents for multi-system workflows.
Status controls
Pause an agent at any time — in-flight operations finish, new ones are rejected with a clear error. Paused agents do not count against your operation quota but do count against the agent limit.