Complete guided setup

The six steps

1. Environment check

Confirms versions, database transaction support, Application Passwords, and whether any Abilities are registered yet.

2. Goal

What you want this site’s agents to do. This shapes the profile suggested in step four; it does not decide anything.

3. First agent

Name, purpose, and environment for one managed identity.

4. What it may do

A suggested starting profile. Read it. Nothing here is applied until you say so.

5. Credential

Opens the Connection Wizard, where you choose a dedicated least-privilege WordPress user and are shown the Application Password once.

6. Finish

Ends with the readiness score for the agent you just created, and what would raise it.

A profile is not a permission

Step four is the step people misread. The suggested profile is a starting point for your decision — it is presented so you have something concrete to accept, narrow, or reject. Setup grants no Ability automatically.

This is intentional. A setup wizard that silently opened permissions would make the first thing the product ever did an unrecorded grant, which is precisely the behaviour the product exists to prevent.

The credential is shown once

The Application Password is revealed a single time, at step five. Copy it into the agent’s configuration before you leave the screen. If you lose it, you do not need to recreate the agent — rotate the credential instead, which keeps the identity and its history intact.

Skipping and resuming

Setup progress is private to the administrator running it, and the flow can be left and resumed. To reopen it later, go to Settings in the plugin’s admin menu. Skipping it entirely is fine — everything it does is available on the individual screens.

Related guides

• SetupVerify the control plane after activation

  Use readiness scoring and runtime verification to test the governance controls on your installed site, rather than assuming the install landed correctly.

• IdentityCreate your first managed agent

  Name an agent for the work it does, tie it to one environment, and see why one agent per purpose is what produces a readable audit trail.

• AuthorityGrant your first permission

  Pick one narrow, reversible, non-public Ability, set it to allow, run one real request, and read the decision it produced before granting anything else.