Semogram Docs
Workspace management

Workspace management

Manage people, integrations, write controls and usage across a workspace

A workspace is the shared boundary for people, projects, plugin installations, data endpoints and administrative settings. Projects organize pipelines, ontology, queries and predictions within that boundary. An endpoint used by two projects is still one workspace resource; granting a project does not create a private copy of it.

You need a Semogram account with access to the workspace. Owners and administrators manage most workspace settings; Company SSO, workspace renaming and workspace deletion require an owner. API-key administration uses the workspace-wide org:manage scope.

Signing in or presenting a key identifies the caller. Operation permissions, workspace and project boundaries, and applicable data grants and write policies are checked separately. Passing an earlier layer does not bypass a later one.IdentityMember account / service API keyOperationRole, API scopes and explicit permissionsResource boundaryWorkspace, project and shared-resource reachData controlsBinding grants / approved write policy and actor grant
Each applicable layer must pass; authentication alone does not grant data access

What you manage

AreaControlsGuide
Workspace identityName, slug, projects and shared resourcesWorkspace and projects
PeopleExisting membership, roles and explicit grantsMembers and roles
Company sign-inVerified domain, SAML connection and allowed employeesCompany SSO
Service accessAPI scopes, project allowlist, workspace resources, expiry and revocationAPI keys
Data readsNamed grants required by ontology read bindingsRead permissions
Data changesPolicy revisions, approval, binding and actor permissionsWrite policies
Consumer interfacesSPARQL mode and governed source-write modeConsumer access
Traffic and spendingRequest windows, usage charges and budget alertsLimits, Usage
InvestigationQuery history, operation charges and decision historyActivity

Access has several layers

A person authenticates with an account; a service authenticates with an API key. Authentication identifies the caller. Authorization then checks the applicable workspace role or operation scope, project/resource boundary and data-specific grants. Writes also need usable policies and supported connector semantics.

For example, a service can have queries:execute for the Maintenance project but still be unable to read an equipment binding that requires maintenance:read. Adding that named grant does not give it permission to edit queries. Selecting Maintenance alone also does not allow it to administer shared plugin installations.

Do not use Full access as a substitute for diagnosing the missing layer. Start with the exact failed operation and resource, then inspect the corresponding controls.

First administration task

Create a dedicated read integration for Maintenance. Select only that project, choose the scopes for its published query, add the actual binding grant and leave unrelated write/manage scopes off. Verify an allowed read and a denied out-of-scope read using a bounded fixture. Retain the key ID and service name; store the one-time secret in your service's secret manager.

The API key guide walks through this complete example. The access model explains each decision layer before you change permissions.