Roles per project, single sign-on and keys that never outgrow their owner, with an audit log of every change by people and agents alike. One console runs users, tenants, features and limits.
Viewer, Member, Admin and Owner. A project is visible to its members only, and the role decides what each of them may do there. Agents join projects just like people.
Your team signs in with Microsoft 365, Google, GitHub or any OpenID Connect provider. Per provider, you decide whether newcomers get an account and which tenant they join.
Every sign-in is listed with its browser and address. Revoke one and its open connection drops at once. A sign-in from an unfamiliar device sends an email alert.
An API key acts as the person who made it, never more. Confine it to chosen projects and everything else is out of reach. It expires after the period you choose.
Every change by a person or an agent is recorded: who, what, when and from which address. Filter the log, link to a single entry, export it as CSV or JSON.
Run several organizations on one installation, each with its own projects, members, sign-in providers and audit log. Switch features off and set ceilings per installation, tenant or project.
Agents connect over MCP with their own account or key, so the same roles bind them. They tell you who is in a project and what has happened there. Adding, inviting and removing people stays with a person.
Say to your agentManager: Who is in the Operations project, which role does each person have, and who is responsible for what?
The agent lists every member with their role and responsibilities note, agents included, so you see at a glance who has access.
Say to your agentManager: Note on Alex's membership that she owns the release process and the on-call rotation.
The agent writes the note on her membership. Her role stays as it is: roles and memberships are changed in the app, by a person.
Say to your agentManager: What has happened in Operations recently, and who did it?
The agent reads the project's recent activity, the trail its dashboard shows, and sums it up by person. Agents appear there under their own names.
Say to your agentManager: Which projects can you reach with this key?
With a key confined to one project, the agent finds that project and no other. The rest is not hidden from it: for that key, it does not exist.
What is there, grouped for the checklist you are working through.
Guides that put this feature to work in a real project.
Every person and every agent works under their own name, in a role you chose, and leaves a record. That holds from a single desktop install to many tenants on your own server.
Get early access