WordPress security, watched and fixed by your AI
Install the free Security Guardian plugin and ask - “is my site safe?”, “find the spam and delete it”, “how did this casino post get here?”, “remove that unknown admin” - the AI finds the problem, explains it, and fixes it with your permission.
- Real-time alerts the moment something changes
- Fixes only with your permission - nothing deleted for good
- Monitoring on/off per site, any time
wpsec_investigate_content3 credits"Best online casino bonus" (risk 92/100) was published by "wp_support", an administrator created 3 days ago with a throw-away email - right after 140 failed logins and a sign-in from a new address. Most likely an attacker created that account.
| Finding | Severity |
|---|---|
| Casino spam post | High |
| New unknown admin | High |
| Plugin with known flaw | Critical |
Shall I revoke "wp_support", delete the post (it goes to the trash) and switch off the vulnerable plugin it probably came through?
https://api.500toolbox.com/api/v1/wpsecurity/mcpClick "Open Claude connectors" below - claude.ai opens on Settings → Connectors.
Click "Add custom connector". Name: WordPress Security Guardian. URL: use the Copy button above and paste it.
Click Add, then Connect. Sign in to 500Toolbox and click Allow. Done.
In a chat, open the tools menu (the + / sliders icon) and switch WordPress Security Guardian on.
A connector added on claude.ai also appears in the Claude desktop and mobile apps. On Team and Enterprise plans an owner adds it once under Organization settings → Connectors; members then click Connect.
Why it is different
Finds it, explains it, fixes it
Watches your site in real time
New posts and comments, new admins, role changes, plugin installs, login attacks, setting and file changes are reported the moment they happen - and spam is checked on arrival.
Smart, not trigger-happy
An article about crypto or gambling law is not spam. It looks for promotion, deceptive links, hidden text and injected code - and tells you exactly why.
Fixes on request
Block attacking IPs, lock out or log out unwanted users, require two-step verification, delete spam, switch off vulnerable plugins, lock the site down - all reversible, and only as far as you allow.
How it works
Protected in three minutes
Install the Security Guardian plugin
Upload the free plugin, open wp-admin → Security Guardian and click Connect - approve on 500Toolbox. Monitoring starts immediately.
Add one URL to your AI app
Paste the server URL as a custom connector in Claude, ChatGPT, Cursor or VS Code.
Ask, then approve
The AI scans, explains what it found, and fixes what you confirm. Deletes wait for your approval on this site.
Free plugin · required
Install the 500Toolbox Security Guardian plugin
The plugin watches your site and reports security events to 500Toolbox in real time, lets scans read your posts and comments, and - only if you allow it - applies the fixes you or your AI ask for. It is separate from the 500Toolbox Connector plugin; you can use either or both.
- Download the plugin zip below (free).
- In wp-admin go to Plugins → Add New → Upload Plugin, choose the zip, then Install Now and Activate.
- Open wp-admin → Security Guardian and click "Connect to 500Toolbox" - approve on the page that opens. Monitoring starts at once.
- In Security Guardian → Settings choose what the AI may do: Monitor only, Approval required or Auto-fix. Login lockout, IP blocking and two-step verification work on the site straight away.
Keep it active: monitoring and fixes run through the plugin. To stop, switch monitoring off here or click Disconnect in wp-admin → Security Guardian. Deleting the plugin removes its key, settings and queue.
What you can ask
Security work, in plain language
“Is my WordPress site safe? Run a full security check.”
“Find spam posts and comments and delete the ones that are clearly spam.”
“Somebody posted casino links on my blog - how did that happen?”
“Which plugins have known vulnerabilities, and which should I switch off?”
“There is an admin called "wp_support" I do not know - revoke it.”
“My site is about gambling law - stop flagging the word "casino".”
“What happened on my site this week?”
“Turn monitoring off for the test site to save credits.”
“Which addresses tried to log in more than 20 times today? Block them for a week.”
“Show me the last 30 days of logins of the user "wp_support", then log them out everywhere.”
“Make the timeline of everything that happened in the last 24 hours.”
“I think someone got in - start an emergency lockdown.”
36 tools
Everything your assistant can do to protect WordPress
Scans, the vulnerability check, investigation, IP blocking, login and user access control, emergency lockdown, fixes, monitoring and lists. Tool names start with wpsec_.
Scan & investigate
Posts, pages, comments and media checked for spam and malicious links, plus the whole health check - one report.
wpsec_scan_siteCasino, crypto-scam, phishing, adult, pharma and SEO spam, hidden links, injected scripts - with reasons and a recommended action.
wpsec_scan_contentVersions and pending updates, admins, risky settings, PHP files in uploads - plus the findings of your last vulnerability check.
wpsec_get_site_healthWordPress, plugins and themes against 44,000+ known vulnerabilities (Wordfence), with the version that fixes each. Only on request, 2 per site per day; queued when our server is busy.
wpsec_check_vulnerabilitiesThe latest check - waiting, running or done - with every vulnerable item and its fix, and the checks left today.
wpsec_get_vulnerability_reportAuthor, account age, revisions, sign-ins and what monitoring saw - the likely cause and the steps to take.
wpsec_investigate_contentPaste a draft, comment or snippet and get the same spam and link analysis.
wpsec_analyze_contentAdmins, editors and authors with sessions and application passwords; new admins and throw-away emails flagged.
wpsec_list_usersFirewall, malware & backups
Mode, rules, rate limits, country rule and the latest blocked or recorded attacks.
wpsec_get_firewallLearning or blocking, switch rules on or off, rate limits for login, XML-RPC, REST and search, country rules.
wpsec_configure_firewallCore and wordpress.org plugins checked against official checksums; themes, uploads and other code against malware patterns.
wpsec_run_malware_scanChanged core files, modified plugin files, web shells, PHP in uploads - and the quarantine.
wpsec_get_malware_resultsMove a malicious file into quarantine (it can no longer run), restore it, or repair a core file from wordpress.org - always with your approval.
wpsec_manage_fileDatabase, files and automatic copies before plugin and theme updates.
wpsec_list_backupsBack up the database or files, schedule daily or weekly backups, restore (with a safety backup first).
wpsec_manage_backupIPs, logins & user access
Logins, wrong passwords, lockouts and refused logins with address and browser, and the most active attacking addresses.
wpsec_get_login_attemptsBlock an address or range for the whole site or just the login, for minutes or for good; unblock; or always allow. Your own addresses can never be blocked.
wpsec_manage_ipEvery rule with expiry, reason and how many requests it stopped.
wpsec_get_ip_rulesBlock / unblock, log out everywhere, remove app passwords, force a new password, require two-step verification, change role - each separate and reversible.
wpsec_manage_user_accessOne step when an intruder may be in: nobody but your protected accounts can add admins, install plugins or create app passwords.
wpsec_emergency_lockdownAlerts, logins and fixes in order, with events close together grouped - what happened, step by step.
wpsec_get_incident_timelineFix (with your permission)
Moves spam to the WordPress trash (restorable 30 days). Needs your approval unless you allow the AI.
wpsec_delete_contentTakes a harmful post offline at once (draft); restore puts it back exactly.
wpsec_quarantine_contentMark spam, hold, approve or trash a comment.
wpsec_moderate_commentSigns them out everywhere, removes their application passwords and lowers them to Subscriber. Their posts stay.
wpsec_revoke_user_accessDeactivates a vulnerable or unwanted plugin; it can be switched on again.
wpsec_deactivate_pluginUndo a quarantine or take a post out of the trash.
wpsec_restore_contentGive a revoked user their old role back.
wpsec_restore_user_accessReal-time monitoring
Status, open alerts, pending approvals, quarantined items and monitoring at a glance.
wpsec_get_security_dashboardSpam, new admins, role changes, plugin installs, login attacks, setting and file changes - as they happen.
wpsec_get_security_alertsRunning, paused or off per site, daily cost and days of credit left.
wpsec_get_monitoring_statusMark an alert dealt with or a false alarm.
wpsec_resolve_alertEvery scan, fix and approval, who asked for it and what changed.
wpsec_get_activity_logLists, policy & sites
Blocked keywords and domains; allowed domains, words and trusted authors to stop false alarms.
wpsec_manage_listsWhen to review or quarantine, and monitoring on or off.
wpsec_configure_policyEvery site with monitoring state and whether fixes are allowed.
wpsec_list_sitesGuided workflows
Built-in prompts
Apps that support MCP prompts (Claude shows them in the + menu) get one-click workflows: a full check-up, a spam clean-up, and "how did this happen?".
Full security check-up
Dashboard plus full scan, explained most urgent first - nothing changed until you say so.
Find and clean spam
Scans comments and posts, shows the reasons, then cleans what you confirm.
How did this spam get on my site?
Finds the worst post and explains how it got there and what to do.
Safety by design
Powerful, with guard rails
You set what the AI may do
Monitor only, approval required, or auto-fix - chosen by you in wp-admin, never by the AI. Anything touching an administrator always waits for your OK.
Everything can be undone
Quarantine, trash, user revocation and plugin switch-off are all reversible. Nothing is deleted permanently and no account is deleted.
You approve deletions
An AI can only request deleting a post or page; you approve it here - unless you choose to let it delete confirmed spam.
Signed and private
Every message between the plugin and 500Toolbox is signed with a per-site key, stored encrypted. Email addresses and passwords never leave your site.
Credits
Pay for what runs - switch off any time
Monitoring costs 2 credits per site per active day; scans and fixes cost by the work they do. Dashboards, alerts and logs are free, failed calls are refunded, and monitoring pauses instead of failing when credits run out.
FAQ
Questions people ask
Is there a firewall?
Yes. Every visitor request is checked for SQL injection, cross-site scripting, path traversal, PHP code injection and scanner probes, with rate limits for the login, XML-RPC, the REST API and search, and optional country rules (behind Cloudflare). Start in learning mode - it records what it would block - then switch blocking on; any rule that misfires can be switched off. Administrators and editors are never checked. Honest limit: a plugin runs after the web server, so a server or CDN firewall stops attacks earlier; this is the layer for sites without one, and a second layer for those with one.
Does it find malware?
It compares WordPress core and every wordpress.org plugin with their official checksums, so changed or added files stand out, and checks themes, premium plugins, must-use plugins and uploads for well-known malware patterns (encoded eval, web shells, request data run as code, PHP hidden in images or uploads). Files can be quarantined (they stop running and can be restored) and changed core files repaired from wordpress.org - always with your approval. Findings are evidence, not proof: a pattern can match legitimate code.
Does it back up my site?
It backs up the database and plugins/themes (uploads optional) on your server, daily or weekly if you like, and copies a plugin or theme automatically before WordPress updates it. Restoring a database first makes a fresh backup, so even a restore can be undone. It is a safety net on the same server - download important backups and keep your host's backups too.
Can I get alerts and reports?
Serious alerts can go straight to Slack or Discord, with a daily or weekly summary of all your sites; your own systems can receive them as signed webhooks. Download a PDF or CSV report per site, or share a read-only report link with a client (revocable at any time). Email alerts are not available yet.
Can it block IP addresses?
Yes. You or your AI can block an address or a range (like 203.0.113.0/24) for the whole site or only the login page, for 15 minutes up to permanently, unblock it, or put it on an always-allowed list. Repeated login lockouts can block an address from the login for 24 hours automatically. Behind Cloudflare the real visitor address is read safely, and addresses you yourself used recently can never be blocked, so you cannot lock yourself out.
What about brute-force attacks on the login?
Smart lockout counts wrong passwords per address and username, so one person mistyping does not lock out a whole office or school on the same network; an address that tries many usernames is locked out of the login as a whole. You choose 3, 5, 10 or 20 attempts and 15 minutes to 24 hours. Two-step verification with any authenticator app, forced password resets and hidden usernames are built in too.
What can I control per user?
Block and unblock, log out of every device, remove application passwords, require a new password, require two-step verification, change the role - each as a separate step you can undo. Your protected accounts (at least the admin who connected the plugin) can never be changed, and the last administrator can never be locked out.
How much may the AI do?
You decide in wp-admin → Security Guardian → Settings, never the AI: "Monitor only" (it scans, explains and recommends), "Approval required" (every fix waits for your OK on this page) or "Auto-fix" (fixes run at once). Anything that touches an administrator, or makes someone one, always waits for your approval.
Does it conflict with Wordfence, Solid Security or other security plugins?
It is built to sit next to them. If another plugin already limits logins or provides two-step verification, the Guardian leaves that part to it automatically (you can still switch it on) so people are not locked out twice. Everything it stores is prefixed, its styles load only on its own page, it never edits .htaccess or wp-config.php, and monitoring, scans and the IP manager work alongside any of them.
Can other plugins or admins switch it off?
With tamper protection on (the default), only your protected accounts can deactivate the Guardian; anyone else trying is stopped and you get an alert, and if it is ever deactivated you are told at once. You can also install a must-use loader so it cannot be switched off from the Plugins screen at all. Honest limit: someone with access to your server files can remove any plugin - that is why the alert matters.
What do the credits pay for?
Two things. 1) Monitoring: while it is on, each protected site costs 2 credits per day that it reports activity - that pays for receiving every event, checking new posts and comments for spam the moment they appear, and keeping the alert history. 2) Tools you or your AI run: a full scan 6, a content scan, investigation or vulnerability check 3, a health check 2, a fix 1; dashboards, alerts and logs are free. Bigger jobs cost more because they use more of our servers. Failed calls are refunded automatically.
What happens when my credits run out?
Monitoring pauses by itself - nothing breaks and no data is lost - and wp-admin shows a notice. As soon as you add credits it starts again within an hour, without you doing anything. Tools that cost credits are refused with a clear message; free tools keep working.
Can I switch monitoring off?
Yes, any time - per site, on this page or in wp-admin → Security Guardian. While it is off nothing is sent and no monitoring credits are used. Scans you ask for still work.
Can the AI really delete spam and remove users?
Yes, as far as you allow in the plugin (Approval required or Auto-fix). It can quarantine or trash spam posts, clean comments, switch off a plugin, and revoke a user (signed out, application passwords removed, lowered to Subscriber). Deleting a post or page needs your approval here unless you also allow the AI to delete confirmed spam. Nothing is ever deleted permanently, no user account is deleted, and your own admin account can never be touched.
Will it flag my legitimate articles?
It is built not to. A word like "casino" or "crypto" alone scores almost nothing; what counts is promotion ("free spins", "guaranteed returns"), suspicious links, hidden text and injected code. Long explanatory articles are scored down. A keyword match alone can never lead to quarantine, and you can add allowed words, domains and trusted authors.
How are malicious links detected?
By their form and context - shorteners, redirectors, bare IP addresses, look-alike domains, throw-away domains, link text that shows one site but goes to another - plus your own block list. Links are never opened. A link no threat database knows is labelled "unverified", never "malicious".
Where do the vulnerability results come from?
From the Wordfence Intelligence database of more than 44,000 WordPress core, plugin and theme vulnerabilities, matched against the exact versions installed on your site. A check runs only when you or your AI ask for one - up to 2 per site per day, 3 credits each. If our server is busy at that moment, the check waits in a short queue and runs by itself, usually within minutes; you pay only when it runs, and a check that fails is free and does not count. Reading the latest result is always free.
Do I still need the WordPress MCP connector?
No. The Security Guardian is its own connector with its own plugin; they work side by side and can be installed, updated or removed independently. Use the WordPress connector for content, users, WooCommerce and courses, and the Security Guardian for protection.
Does the plugin slow my site?
No. Events are queued and sent in the background a few seconds later; file checks run twice a day. Visitors never wait for 500Toolbox.
What is sent to 500Toolbox?
Security events and, if you allow content checks, the text of posts and comments so they can be scanned. Email addresses are never sent (only their domain), and passwords, cookies and keys are never read. You can switch content checks off in the plugin.
Your WordPress site, guarded around the clock
Install the plugin once - your AI keeps watch from then on.