Roles & Permissions
Menatic AI uses role-based access control (RBAC) to restrict what users can see and do, with permissions grouped by resource type and action.
Built-in Roles
Section titled “Built-in Roles”| Role | Description |
|---|---|
admin | Full access to all resources and settings |
user | Basic access — read notebooks and models |
You can create custom roles with any combination of permissions.
Permission Structure
Section titled “Permission Structure”Permissions follow the pattern <resource>:<action>:
Jupyter Notebooks
Section titled “Jupyter Notebooks”| Permission | Description |
|---|---|
jupyter:read | View all notebooks |
jupyter:read_own | View only own notebooks |
jupyter:write | Create new notebooks |
jupyter:delete | Delete notebooks |
MCP Servers
Section titled “MCP Servers”| Permission | Description |
|---|---|
mcp:read | View all MCP servers |
mcp:write | Deploy new MCP servers |
mcp:delete | Delete MCP servers |
AI Models
Section titled “AI Models”| Permission | Description |
|---|---|
models:read | View available models |
models:write | Add/configure models |
models:delete | Remove models |
Workflows
Section titled “Workflows”| Permission | Description |
|---|---|
workflows:read | View workflows and builds |
workflows:write | Create and run workflows |
workflows:delete | Delete workflows |
Settings
Section titled “Settings”| Permission | Description |
|---|---|
settings:read | View system settings |
settings:write | Modify system settings |
Creating a Custom Role
Section titled “Creating a Custom Role”- Go to Settings → Roles
- Click New Role
- Enter a role name
- Toggle individual permissions on/off
- Click Save
Object-Level Permissions
Section titled “Object-Level Permissions”The read_own action on the jupyter resource enables object-level access control. Users with this permission can only see notebooks where they are assigned as the owner.
This is useful for:
- Multi-tenant environments where users should not see each other’s notebooks
- Teams where each member has their own isolated environment
To use read_own:
- Assign a user the role with
jupyter:read_own(notjupyter:read) - When creating notebooks for this user, set their account as the Owner
Assigning Roles
Section titled “Assigning Roles”Roles are assigned per user in Settings → Users. A user can have one role. To combine permissions, create a custom role that includes the required permissions.