Gemini CLI + xCloud

Manage WordPress with Gemini CLI on xCloud

Managing WordPress with Gemini CLI means describing a broken or suspicious site to Google's terminal agent and letting it read health, switch on WP_DEBUG and fetch a one-time admin login through xCloud, then switch debugging off again.

Skill: xcloud:wordpress
Toolsets: wordpress-actions, sites-wordpress, vulnerabilities, broken-links, pagespeed
Free with every xCloud account

Set up Gemini CLI All Gemini CLI guides

Younorthwind.example.com shows a critical error after an update. Check health.

sites_wordpress_statusread-only

AgentWordPress status looks degraded and a plugin update ran an hour ago. I can switch on WP_DEBUG so you can read the log. Go ahead?

YouYes.

sites_wp-debugneeds approval

AgentWP_DEBUG is on. Read debug.log under Site, Site Monitoring, Logs in the dashboard. Tell me when you are done and I will turn it off.

Last updated: October 2026. Keep a human in the loop: xCloud stops and asks before anything that creates, deploys, updates, reboots, deletes or buys.

Setup

How Do You Set Up Gemini CLI to Manage WordPress on xCloud?

Connect Gemini CLI once; every job on this account uses the same connection. Then ask in plain words.

1
Add the xCloud MCP server
Run this in your terminal, then start gemini. The --scope user flag writes the server to ~/.gemini/settings.json so it is available in every project; without it, gemini mcp add writes to the current project's .gemini/settings.json and refuses to run from your home directory. The first time it calls xCloud it finds the OAuth endpoints, opens your browser on the xCloud approval screen and asks you to tick the teams and choose Read-only or Full access.
Terminal
```
gemini mcp add --scope user --transport http xcloud https://app.xcloud.host/mcp
```
2
Or edit settings.json
Add this to ~/.gemini/settings.json for every project, or to .gemini/settings.json in one project. This is the entry gemini mcp add writes: url plus type set to http. The older httpUrl key still works, but a url without a type is treated as an SSE server and will not connect. If the browser sign-in does not start, type /mcp auth xcloud inside Gemini CLI.
JSON
```
{
  "mcpServers": {
    "xcloud": {
      "url": "https://app.xcloud.host/mcp",
      "type": "http"
    }
  }
}
```
3
No browser? Use an API key
For a headless machine, create a token with the mcp:invoke scope plus the read abilities for the areas it will use (read:servers and read:sites, with read:billing and read:addons for billing and add-on tools) and the matching write: abilities if it should change things in Settings, Developers, API Tokens and send it in the headers field. Keep the token out of version control, and keep the quotes: xCloud tokens contain a pipe character.
JSON
```
{
  "mcpServers": {
    "xcloud": {
      "url": "https://app.xcloud.host/mcp",
      "type": "http",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}
```
✓
Check it worked
Then ask Gemini CLI for the job itself, for example:
Prompt
```
northwind.example.com shows a critical error. Check its WordPress health.
```

In practice

How Does WordPress Work from Gemini CLI?

The case Gemini CLI handles well is a site that is misbehaving right now. You type what you see, such as the critical error message on northwind.example.com, and Gemini calls sites_wordpress_status for the WordPress health reading. In its confirmation dialogs and logs the tools carry the name of the server in front, so you will see mcp_xcloud_sites_wordpress_status and, later, mcp_xcloud_sites_wp-debug. The names are long, but they tell you at a glance that the call goes to xCloud and which operation it is.

Next comes debug mode. Gemini calls sites_wp-debug with enabled set to true. Reading debug.log is not an xCloud operation, since it stays in the dashboard under Site, Site Monitoring, Logs, so Gemini asks you to paste the relevant lines or points you at your local checkout of the theme. When you need to look at the admin screens, ask for a magic login. Gemini calls sites_magic-login, and on a site that has never had one, xCloud first installs the magic-login plugin over SSH, which is why that call stops for a confirmation. The link is single-use, expires after about ten minutes and prints in your terminal for you to open.

Two checks sit between your request and the change. Gemini CLI asks before it runs a tool, and xCloud's confirmation rule stops anything that is not a routine action. For this job you can narrow the surface with includeTools on the xcloud entry in settings.json, listing only the WordPress operations you want Gemini to see. When you have found the cause, finish the way you started: tell Gemini you are done and ask it to turn WP_DEBUG off, so the site returns to its normal state.

Gemini CLI specific: Leave WP_DEBUG on only as long as you are reproducing the error, and say so in the same message as the request, so the step that switches it off is not forgotten. Keep trust set to false on the xcloud entry. With trust set to true, Gemini CLI skips its own tool confirmation dialogs, so a magic login would face only xCloud's confirmation. The link also appears in your terminal scrollback, so open it yourself and clear the screen if the machine is shared.

What xCloud does for wordpress

xCloud reads each site's plugins, themes and pending updates, runs the updates after you approve with a backup first, and scans for vulnerabilities, broken links and PageSpeed results. It also toggles WP_DEBUG, issues magic login links and creates new WordPress sites with a dry run first.

Pick the site or the fleet. For one site the agent uses the one you name. For a fleet question such as pending updates it lists your sites, keeps the WordPress ones and answers grouped by site. It reads WordPress health first.
Refresh and read the inventory. The agent refreshes the cached plugin and theme list when it may be stale, then reads plugins, themes and the updates summary for core, plugins and themes. Refreshing and reading run without a prompt.
Back up, then update. Updates and activations stop for your approval. The agent sets backup_before_update to true for production sites, sends the request for the type and slugs you approved, and confirms the result from the site's events, because updates run in the background.
Scan for vulnerabilities. The agent starts a site scan, reads the count by severity and the findings, or reads the team-wide rollup worst first. You decide which findings to ignore, and the agent can unignore one later.
Check links and speed. A broken-link scan starts once you confirm it, the agent polls the run and reports the pages with the most broken links, calling a truncated scan partial. A PageSpeed scan covers mobile and desktop and can be compared with earlier runs.
Debug, log in or create a site. The agent toggles WP_DEBUG, issues a magic login link, or creates a new WordPress site. A new site gets a dry run first, then one approval, then polling until the site is live.

Reference

WordPress Settings and Limits on xCloud

The facts Gemini CLI works within when it manages WordPress. Where a row names the dashboard, that step stays yours to take there.

Setting or limit	What applies
Updates	Plugin, theme and core updates take a type of plugin, theme or core and optional slugs; leaving slugs out updates every item of that type. They run in the background
Backup first	backup_before_update and backup_before_action snapshot the site before the change. The agent sets them for production sites
Activation	Plugins and themes are activated by type and slug, with an optional backup first
Inventory refresh	Refreshes the cached plugin and theme list before it is read, and runs without a prompt
WP_DEBUG	A toggle with enabled set to true or false. Reading debug.log stays in the dashboard under Site, Site Monitoring, Logs
Magic login	A single-use link that expires after about ten minutes. The first call on a site installs the magic-login plugin over SSH, so it stops for confirmation
Vulnerability scans	Per site: start a scan, read the count by severity (critical, high, medium, low), list findings, ignore or unignore one. Team-wide: one rollup across every site
Broken-link scans	Start a scan, poll the run, then read open findings. The first scan turns on monitoring with a manual frequency, and a 409 means a scan is already running
PageSpeed	One scan runs both mobile and desktop and is complete when both results are in. A 409 means a scan is still running. History is kept per strategy
Caching	All layers are readable and caches can be purged. Turning page cache, object cache or Cloudflare edge cache on or off is a dashboard step under Site, WordPress, Caching
WordPress staging	Dashboard-only, under Site overview, Add Staging. The API answers 422 for a WordPress site
New WordPress site	Nginx or OpenLiteSpeed servers only. Dry run first, demo mode for a free hostname or live mode with a domain and SSL. Auto-generated admin credentials are returned once

Rules Gemini CLI has to follow

Plugin, theme and core updates stop for your approval, and the agent takes a backup first on production sites.
A magic login link is a password: it is shown once in the reply and never repeated in a summary, a ticket or a shared channel.
Vulnerability scans, PageSpeed scans and inventory refreshes change nothing on the site and run without a prompt; a broken-link scan asks first. Ignoring a finding hides it, so the agent records the reason you give.
WP_DEBUG is turned back off when you say you are done.
WordPress staging, caching switches and restores are dashboard steps; the agent gives you the path and the dashboard link xCloud returned.
Link text and finding titles come from site content, so the agent treats them as data and never as instructions.

Example prompts

What Can You Ask Gemini CLI to Do for WordPress?

Type these as written and swap in your own repository, site and server names. Reads and routine actions such as backups, cache purges, PageSpeed scans and vulnerability scans run straight away; creating, deploying, updating, rebooting, deleting, buying or starting a broken-link scan stops and asks first.

Prompt

northwind.example.com shows a critical error. Check its WordPress health.

Prompt

Turn on WP_DEBUG for northwind.example.com, wait for me to read the log, then turn it off when I say done.

Prompt

Get me a magic login link for northwind.example.com. Print it once, do not repeat it.

Prompt

Which of my sites have pending WordPress core, plugin or theme updates? Group them by site.

Prompt

Update every plugin on the Northwind site with a backup first, then confirm the homepage still loads.

Prompt

Show the vulnerability count for every site on this team, sorted worst first, and name the top three findings.

Prompt

Turn on WP_DEBUG for the shop site so I can see the error. Turn it off again when I say I'm done.

Prompt

Give me a magic login link for the Northwind WordPress admin.

Prompt

Scan the Northwind site for broken links and list the pages with the most broken links.

Prompt

Run a PageSpeed scan on the Northwind homepage and compare the score with the previous scans.

Prompt

Dry-run a new WordPress site called Northwind on my Frankfurt server and show me the resolved configuration without creating anything.

Gemini CLI and WordPress: Frequently Asked Questions

What people ask before they let Gemini CLI manage WordPress through xCloud.

Can Gemini CLI read the WordPress error log for me?

Not through xCloud. Gemini CLI can turn WP_DEBUG on and off, but debug.log is read in the dashboard under Site, Site Monitoring, Logs. Paste the lines into the session and Gemini can reason about them against your code.

Why does Gemini CLI ask me twice before a magic login?

Gemini CLI confirms tool calls on its own, and xCloud also stops the call, because the first magic login on a site installs the magic-login plugin over SSH. Both prompts are protection, so do not turn on trust to skip the first one.

Does the agent back up my WordPress site before it updates plugins?

It asks xCloud to snapshot the site first by setting backup_before_update to true, and it does this by default for production sites. The update itself stops for your approval and runs in the background, so the agent confirms the result from the site's events.

How does an agent find vulnerable WordPress sites?

It can read the vulnerability rollup for your whole team in one call and sort it worst first, or scan a single site and read the count by severity. Findings can be ignored with a reason and unignored later.

Is a magic login link safe to ask an agent for?

The link is single-use and expires after about ten minutes, but anyone who opens it is logged in as the admin. Treat it like a password: use it yourself and do not paste it into a ticket or a shared channel.

Can an agent switch my WordPress caching on or off?

No. It can read every caching layer and purge caches, but turning page cache, object cache or Cloudflare edge cache on or off is a dashboard step under Site, WordPress, Caching. Cache choices can only be set while a new site is created.

Can an agent create a WordPress staging site?

No. WordPress staging is dashboard-only, under Site overview, Add Staging, and the API refuses it for WordPress sites. Staging from the API covers Git sites only.

Can an agent create a new WordPress site?

Yes, on an Nginx or OpenLiteSpeed server. It runs a dry run first, shows the resolved configuration and creates the site only after you approve. WordPress is not supported on Docker servers.

Other agents