A WordPress security audit, the way I do it, is a written review of one site: its plugins and themes, user accounts, WordPress and server configuration, HTTP headers and backups. You get a document that lists what I found, how serious each finding is and how to fix it. I am Mariusz Szatkowski, a WordPress developer since 2006, and I do the audit myself. I do not run a security operations centre, sell round-the-clock monitoring or offer penetration testing as a service, and nobody can honestly promise you a site that will never be hacked.
Who carries out the WordPress security audit?
I do, personally. WPPoland is the name I work under as a senior WordPress developer based in Gdynia, Poland. No team, no subcontractors: the person who reads your code and logs is the one who writes the report.
What does a WordPress security audit include?
- installed plugins and themes checked against known vulnerabilities, plus a read of custom code and abandoned plugins
- user accounts: who has administrator access, application passwords, accounts nobody remembers creating
- WordPress configuration:
wp-config.phpconstants, salts, file editor, XML-RPC, REST user endpoints - server side: file permissions, PHP execution in uploads, HTTP security headers, TLS
- backups: where they are, how long they are kept and whether a restore has ever been tested
- AI plugins that give an external agent code execution, database or file access
- if the site is already compromised: the likely entry point and the malware removal steps
Where is the audit carried out?
Remotely, from Gdynia, for site owners in the UK, Ireland and elsewhere. Communication runs by email and in writing, so every finding and decision can be looked up later.
How much does a WordPress security audit cost?
The price is individual and depends on the number of plugins, the amount of custom code and whether the site is already infected. Send a short description of the site and I reply with the scope and a written quote. Fixes from the report are quoted separately.
How WordPress sites get hacked
Most attacks are automated: bots scan the web for known vulnerable versions and try them on every site they find. The site owner is rarely a chosen target.
WordPress plugin vulnerabilities
The risk sits in plugins. Patchstack counted 11,334 new vulnerabilities in the WordPress ecosystem in 2025, about 91% of them in plugins, 9% in themes and 6 in core (Patchstack, State of WordPress Security in 2026). The shapes repeat:
- REST endpoints registered with
permission_callback => '__return_true', open to anyone - stored XSS through shortcodes or blocks that echo
$_GETor post meta without escaping - file upload handlers that skip
wp_check_filetype_and_ext()and a MIME allowlist - settings saves that trust the request and let a subscriber change options or roles
I check installed slugs and versions against the Patchstack and WPScan databases, then read the source of custom plugins and the theme for the patterns above. A database lookup alone misses code nobody has catalogued. AI-generated code ships these gaps often (missing nonces, missing capability checks, unsanitised input), so a site built with an AI tool may also need the AI-built website rescue work.
Brute force on wp-login.php and XML-RPC
/wp-login.php and /xmlrpc.php see login attempts on most public sites. On XML-RPC, system.multicall lets one request carry many password attempts. If nothing on the site uses XML-RPC, it can be blocked; if Jetpack or the WordPress mobile app needs it, limit it at the firewall instead.
WordPress user enumeration
?author=1 redirects to /author/<slug>/ and reveals a login name. The same comes from /wp-json/wp/v2/users on installs that never restricted the public users endpoint. On a hardened site both return 404 or an empty list.
Plugin supply chain
The realistic risk is an abandoned plugin sold to a new owner who ships a tracker in the next update, or a commercial plugin whose own update server gets compromised. I list where each plugin updates from and flag plugins that have changed hands or stopped receiving updates.
Are AI plugins with code execution a security risk?
A vulnerability scanner looks for bugs: outdated versions, known CVEs, missing validation. A different class of plugin gives an AI agent PHP execution, database queries and file operations on purpose, through MCP, a CLI bridge or a webhook. That is a documented feature, and in effect it is remote code execution for whoever holds the key. A CVE scanner reports such a plugin as clean, because there is no bug to report.
For each such plugin I check:
- whether the endpoint requires authentication and whether it is reachable from the public internet
- where the credentials live (
wp-config.php, the options table, environment variables) and who can read them - whether executed operations are logged in a form you can reconstruct after an incident
My default recommendation is to keep these plugins off production and use them on staging only, behind an IP allowlist.
Hardening recommendations in the report
Hardening means constraints that make the next attack harder to land. These are the changes I most often recommend, roughly in this order.
wp-config.php constants. DISALLOW_FILE_EDIT turns off the code editor in the dashboard, and DISALLOW_FILE_MODS blocks plugin and theme installs where deployments go through Git. FORCE_SSL_ADMIN keeps the admin on HTTPS. The file should be readable by the web server user only where the host allows it.
Secrets. Rotate the eight salts with the wordpress.org generator, which logs everyone out, and revoke application passwords that no documented integration uses.
Login. Two-factor login for administrators (the Two Factor plugin or Wordfence Login Security with TOTP), with a documented way to reset it over WP-CLI when a phone is lost. Login throttling at the edge as well as in PHP, for example a rate limit on POST requests to /wp-login.php.
File system. PHP execution disabled in /wp-content/uploads/ through an .htaccess rule or the nginx equivalent, directory listing off, xmlrpc.php blocked unless the site needs it.
Detection. A regular comparison of core and plugin files with the official checksums (wp core verify-checksums, wp plugin verify-checksums --all), and an alert when a new administrator account appears.
If you order the fixes, I apply them on staging first and then on production.
Breach patterns I look for
The persistent admin user. A vulnerable form plugin lets an attacker create a user through an unprotected AJAX endpoint, with the administrator role and a harmless name like “Support”. Restoring last week’s backup does not help if the account was created earlier. I compare wp_users and wp_usermeta with a known clean state rather than trusting the list in the dashboard.
The uploads backdoor. A PHP file hidden somewhere like /wp-content/uploads/2024/03/.cache.php accepts a payload by POST and runs it. A partial restore of core and plugins leaves it in place. I search uploads for .php, .phtml and .phar files and check that PHP execution is disabled there at server level.
The leaked key. A .env file or a backup of wp-config.php sits in the web root or in a public repository, with SMTP or payment API keys inside. I check for .env, .env.backup, wp-config.php.bak and any file under the web root that contains DB_PASSWORD or _KEY.
What you get after the audit
A written report, sent by email as a document:
- every finding with its priority (critical, high, medium, low), where it was found and the evidence: a log line, a header, a file path, a CVE reference where one exists
- a concrete fix for each finding, written so that your developer, or I, can apply it without guessing
- a short list of what I could not check, and why (for example no server access or no logs)
Questions about the report go by email. If you order the fixes, I check each one again afterwards and mark it as closed in the report.
What access the audit requires
- A temporary WordPress administrator account with two-factor login, created for the audit and removed afterwards.
- Read-only SFTP or SSH access. It lets me verify core files with WP-CLI, check permissions, read
access.loganderror.log, and look for files hidden in uploads. Without it, part of the site stays invisible and the report says so. - A list of integrations and important plugins, a few lines each: what it does and who relies on it.
Google Search Console access and any earlier audit report help too. Before I get access to a site with personal data, I sign a data processing agreement under Art. 28 GDPR.
Hacked website: what to do in the first hour
- Delete nothing. Copy the files and the database as they are. Infected files are evidence of how the attacker got in.
- Change passwords from a clean device: WordPress admin, hosting, SFTP, database, payment and mail API keys. Log out every session.
- Take the site offline or enable maintenance mode if it serves malware, redirects visitors or collects form and payment data.
- Ask the host for
access.loganderror.logand how long they keep them. Without logs nobody can tell what was read. - Check Google Search Console. An unknown owner is a common sign of a break-in; remove it once the copy is safe.
- Write down when you confirmed the breach. The notification deadline in the next section counts from there.
How does a hack cleanup work?
I work on a copy first. I look for the entry point (logs, wp core verify-checksums, a diff of plugin and theme files), remove malware and backdoors from files and database, check administrator accounts and cron jobs, reinstall core and plugins from clean sources and rotate every secret. If Google flagged the site, you request a review in Search Console once it is clean. How long it takes depends on how much is infected and whether logs exist, so I estimate only after looking at the site. You get a written summary with what I found about possible data exposure.
Do you have to report a data breach after a hack?
You are the controller, so the decision to notify is yours and your data protection officer’s, if you have one. My part is the facts a lawyer needs: whether personal data sat in the database or files, whether the attacker could reach it and since when.
- Ireland and the rest of the EU. Article 33 GDPR requires the controller to notify the supervisory authority without undue delay and, where feasible, within 72 hours of becoming aware of the breach, unless it is unlikely to result in a risk to people’s rights and freedoms. In Ireland that authority is the Data Protection Commission, established by section 10 of the Data Protection Act 2018. Where the risk to people is high, Article 34 also requires telling them. Source: GDPR on EUR-Lex.
- United Kingdom. The UK GDPR keeps the same rule in Article 33: notify the UK regulator without undue delay and, where feasible, within 72 hours, and document every breach, reported or not.
When does a WordPress site need a security audit?
When at least one of these is true: the site takes payments or stores personal data, nobody has rotated the salts or reviewed application passwords in the last 12 months, you took the site over from another developer or agency, or something looks wrong (admin emails you did not expect, search results showing pages you did not write, outbound mail flagged as spam).
If none of that applies and the site is a small brochure site on managed hosting with updates running, an audit can probably wait, and I will tell you so in my reply.
Send a short description through the contact form: the address, what the site handles and anything that worries you. Before that, you can look at the public side yourself: the WordPress site check fetches one address and reports measured numbers, with no email address required. Ongoing care after an audit is described under WordPress maintenance.




