Run Your First Operation

This is the moment everything comes together. You’ll send your agent a real command, watch it call Maxi AI Core on your site, and confirm the result.

Before you begin: make sure you’ve completed Connect your AI and your agent is configured.

What “Operation” Means

In Maxi AI Core, an operation is anything the agent does on your site through an MCP ability — a query, a content edit, a setting change, a background job. Every message you send your agent ultimately turns into one or more operations behind the scenes.

The point of this section is to issue a real one and see it work end-to-end.

Step 1 — Pick a Safe First Command

Start with a read-only command. There’s nothing to break and nothing to undo. Try any of these by simply asking your agent in plain language:

“What version of Maxi AI Core is running on my site?”

Exercises maxi/get-site-info — returns the plugin version, license status, and the feature groups your license unlocks.

“List my 5 most recent posts.”

Exercises maxi/list-content — returns titles, IDs, statuses, and slugs.

“Show me the title and status of post ID 1.”

Exercises maxi/get-content — returns the full record of a single post.

“Show me my license details.”

Exercises maxi/get-site-info again, this time focused on the licensing fields.

Pick one and send it.

Step 2 — What You’ll See on the First Turn

The very first message in a new agent session has a hidden first step. Before answering you, the agent calls maxi/bootstrap-session, which:

• Loads the operational playbook (the rules the agent must follow).
• Loads any operator notes you’ve written (your authoritative instructions to the agent).
• Loads available knowledge notes (solutions captured from past sessions).

This happens silently — you’ll just see the answer to your question. From the second turn onward, the agent goes straight to the ability you asked about.

Step 3 — Try a Small Write

Once a read works, try a write to see the full mutation flow:

“Create a draft post titled ‘Hello from Maxi’.”

This calls maxi/create-content. On the first write of any session, you may see the agent briefly stop, read a rule, then retry — that’s the rules handshake in action. It’s normal, and it only happens once per ability per session.

The result is a draft post in your site. Nothing is published, nothing is visible to visitors, and you can delete it whenever you want.

Step 4 — Verify the Result

Two ways to confirm the operation actually happened:

Spot-check in wp-admin.

• For the read commands above, open wp-admin and compare what the agent told you against what you see — the version number on Settings → Maxi AI License, the post titles under Posts → All Posts, and so on.
• For the draft post, go to Posts → All Posts and look for “Hello from Maxi”. You can open it, edit it, or move it to trash like any normal post.

Ask the agent for the audit log.

“Show me last 10 audit log events.”

This calls maxi/get-audit-events and lists every ability call made on your site, the user it ran as, the timestamp, and whether it succeeded. This is the authoritative record — if an operation is in the audit log, it really happened.

If Something Didn’t Work

The most common first-run issues — 401 Unauthorized, rules_not_acknowledged, write_not_authorized, license_required — are covered in the Connect your AI → Troubleshooting section.

What’s Next

You now have a working agent that can read your site, write to it, and produce a verifiable trail of every action. From here, the rest of the documentation covers what each ability does in depth, how to write operator notes that shape your agent’s behaviour, and how to build playbooks for repeatable multi-step tasks.