Dashboard
Settings, roles & memory
Organization controls, access roles, and project memory modes.
The Settings screen controls organization- and project-level preferences: who can do what, how project memory behaves, and optional usage caps.
Roles#
Access is role-based. Capabilities gate sensitive actions like editing policy or deleting a project.
| Role | Typical access |
|---|---|
| Admin | Full control, including settings, roles, and deletion. |
| Security reviewer | Configure policy and review traces. |
| Developer | Integrate and view telemetry. |
| Viewer | Read-only access to dashboards. |
Memory modes#
Project memory stores human-confirmed context for repo-backed review projects. The mode sets whether an agent may propose or persist that context. It does not train classifiers or tune gateway thresholds. See Thresholds and calibration and Tenant adaptation for current rollout status.
| Mode | Behavior |
|---|---|
off | No persistent memory actions. |
propose_only | Show proposals without writing. |
confirmed_writes | Allow human-confirmed persistence. Required to edit AI review scope and context. |
Capacity planning#
A project can store an optional expected monthly volume (checks_per_month), edited on the Policy configuration screen’s General tab. The field is capacity-planning metadata. It does not reject requests or act as a quota.