Skip to content

Permissions

Permissions decide what a client can and cannot do on a hub.

The hub answers where does the client connect? The client answers what is connecting? Permissions answer what is allowed?

Create permissions after the client exists and the hub is clear. Use Live Map later to confirm the connected client behaves as expected.

  • access is too wide or too narrow;
  • a client sees permission denied;
  • a client action is access blocked;
  • a client can do too much or too little;
  • a permission set is hard to review;
  • you need to split broad access into smaller purpose-based rules.
QuestionSimple answer
What are permissions?Rules for what a client may do on a hub.
When do I create them?After the client exists.
What should they include?The hub, linked clients, allowed actions, blocked actions, and notes.
What is safest first?Narrow access for one purpose.
What should be rare?Everything access or full control.
Permission review screen showing purpose, linked clients, allowed actions, blocked actions, and review state.
Good permissions are readable before you open every detail: purpose, clients, allowed actions, and blocked actions should line up.

Permissions usually include:

  • a hub it belongs to;
  • one or more clients;
  • allowed actions;
  • blocked actions;
  • optional safety rules;
  • a description for team context;
  • notes that explain why the access exists.
SpecificSafety rulesEverything access

Specific permissions allow only named actions. Use these for most clients.

Safety rules keep permissions intentionally narrow and safer for public-facing clients.

Everything access is wide. Use it only when the client has a trusted purpose and the risk is understood.

Read permissions from top to bottom:

  1. Hub: the place these permissions apply.
  2. Linked clients: the clients that receive these permissions.
  3. Allowed actions: what those clients can do.
  4. Blocked actions: what stays blocked.
  5. Notes: why these permissions exist.

If the client name and permissions name do not clearly explain the relationship, rename or describe them before adding more rules.

Create permissions after the client exists. That makes the relationship easier to read.

  1. Open Permissions. Select Create.
  2. Choose a hub. Permissions always apply to one hub.
  3. Name the permissions. Use a purpose-based name, such as frontdesk-questions-only.
  4. Link clients. Select the clients that should receive this access.
  5. Choose the shape. Use narrow access for most clients.
  6. Review allowed and blocked actions. Allow only what is needed and block risky actions.

Before saving, read the permissions sentence in plain language:

These clients can do these actions on this hub, and these actions stay blocked.

If that sentence is hard to say, split the permissions or rename them before adding more clients.

For public hub access, Thalovant favors narrow defaults:

  • no wider message sending;
  • no escalation, which means no higher-level requests;
  • no propagation, which means no passing work along;
  • no full-control access;
  • normal question access only.

That keeps public clients useful while limiting accidental wide control.

If access feels too wide, too narrow, or the client can do too much or too little, start by reading the permission shape before changing allowed actions.

If someone describes the problem as too much access, use this section before adding another blocked action.

If the visible message says permission denied or access blocked, compare the linked client, allowed actions, and blocked actions before changing the hub.

StyleUse it whenTypical access
Question-onlyA user-facing client should ask normal questions.Questions and status, with wider controls blocked.
ObserverA client only needs health or status.Read-only access.
Skill helperA client helps Thalovant see available skills.Skill sharing plus only the small reads it needs.
Temporary taskA trusted client performs one narrow job.Specific actions with a clear description.
Permission denied state showing linked client, allowed question action, and blocked runtime update action.
Permission denied is easier to fix when the linked client, allowed action, and blocked action are visible together.

When a hub has a client sharing live skill information, the permissions builder can show available skills. This helps you block exact items without guessing names.

If live suggestions are not available yet, permissions still work. Start simple, then return later when suggestions appear.

Permissions are ready when:

  • the name explains the purpose;
  • the linked clients all need the same kind of access;
  • allowed actions are specific;
  • wider controls are blocked unless there is a written reason;
  • notes explain why the access exists;
  • a teammate could read the permissions sentence without guessing.