Agents
Admin UI Prompt
Use this prompt when designing customer-facing controls for Neutron AI memory inside your own product.
Design a clean admin experience for managing Neutron AI memory in a multi-tenant product.
Audience: product admins, support leads, AI operations teams, and developers integrating agent memory.
Primary jobs:
- View memory by tenant, workspace, project, user, and agent scope.
- Search saved memory by type, source label, and date.
- Inspect a memory record without exposing fields the current user is not allowed to see.
- Delete incorrect, stale, or sensitive memory.
- Configure which agents can access which scopes.
- Review safe audit events for memory reads, writes, context generation, and deletes.
- Test SDK, API, CLI, and MCP connectivity.
UI principles:
- Use plain language such as saved memory, scopes, agents, and connectors.
- Make access scope visible on every detail view.
- Show loading, empty, error, and success states.
- Keep destructive actions explicit and reversible only when your product policy allows it.
- Do not expose hosting providers, private topology, runtime details, resource names, or operational runbooks.
The admin UI should help customers control memory safely. It should not expose private operational details.