wp2shell-detect

Blackbox, non-intrusive detector for wp2shell — the pre-authentication RCE chain in WordPress core disclosed on 2026-07-17:

• CVE-2026-63030 — route confusion in the REST batch endpoint (/wp-json/batch/v1): a sub-request whose path fails wp_parse_url() lands in the validation array but not the handler array, desynchronising the two.
• CVE-2026-60137 — SQL injection in WP_Query's author__not_in handling, reachable pre-auth through that route confusion.

Chained, they let an anonymous attacker forge an administrator account and drop a plugin webshell — on a default install, no plugin or theme required.

Full technical writeup: https://own2pwn.fr/articles/appsec/wp2shell-wordpress-rce

What this tool does (and does not)

It detects exposure. It does not exploit anything. Every request it makes is a normal, read-only request a crawler could send:

1. Fingerprints the WordPress core version from many public sources, so a hardened install that strips the generator tag is still identified: the generator meta tag, core asset ?ver= (on /wp-includes/ and /wp-admin/ assets, which track core), the RSS feed generator (?v=), /wp-links-opml.php, the login page assets, /wp-includes/version.php, readme.html, and the /wp-json/ index. Sources are cross-checked and a confidence (high when ≥2 agree) is reported.
2. Compares the version to the known-vulnerable ranges.
3. Checks the REST batch route with a benign OPTIONS request (and the REST index) — which returns the route schema without ever invoking the handler — to tell whether an edge mitigation is in place.
4. Discovers headless WordPress back-ends. If the target front isn't WP but references a WordPress host on the same domain (via CSP, the Link header or asset URLs — common with headless/Next.js fronts), the tool reports it, and --discover scans it too.

No request triggers the vulnerability. No POST /wp-json/batch/v1, no author_exclude, no SQL. If the core version is hidden, the tool says so rather than guessing (confirming wp2shell blindly would require exploiting it).

Resilience: retries with backoff, gzip/deflate decoding, https→http fallback, cross-host redirect following, a browser-like User-Agent (overridable), and parallel multi-target scanning.

Vulnerable ranges

Install

pip install wp2shell-detect

Or run it straight from source (single file, no dependencies, Python 3.8+):

git clone https://github.com/own2pwn-fr/wp2shell-detect
./wp2shell-detect/wp2shell_detect.py https://example.com

Usage

Zero dependencies. Python 3.8+.

wp2shell-detect https://example.com
wp2shell-detect --json https://a.example https://b.example
wp2shell-detect --targets hosts.txt --workers 16   # parallel sweep
wp2shell-detect --discover https://front.example   # follow headless WP

(Installed from PyPI, the command is wp2shell-detect. From a clone, call ./wp2shell_detect.py with the same arguments.)

Options: --json, --discover, --workers N, --timeout S, --user-agent UA, --insecure, --targets FILE. Exit code 2 if any target is found vulnerable (handy in CI / cron sweeps).

Headless WordPress

Modern sites often serve a JS/Next.js front on www. and keep WordPress on a sibling host (wordpress., wp., cms.). Scanning www. then returns not WordPress — correctly, because that host isn't WordPress. Point the scan at the WordPress host, or pass --discover to have the tool find and scan it from the front's CSP / Link header / asset URLs automatically.

Example:

[!!] https://blog.example.com  ->  VULNERABLE (pre-auth RCE)
      version : 6.9.3  (generator meta tag=6.9.3, RSS feed generator=6.9.3)
      batch   : reachable  (/wp-json/batch/v1)
      cves    : CVE-2026-63030, CVE-2026-60137
      note    : WordPress 6.9.3 is in the wp2shell pre-auth RCE range. REST batch endpoint reachable — no edge mitigation.

Mitigation

Patch WordPress core to 6.8.6 / 6.9.5 / 7.0.2 or later. If patching must wait, block the batch route at the edge (/wp-json/batch/v1 and ?rest_route=/batch/v1) or via a must-use plugin hooking rest_pre_dispatch, then audit for rogue admin accounts and unknown plugins.

Legal

For authorised assessment only. Running this against systems you neither own nor are mandated to test may be illegal in your jurisdiction. You are responsible for how you use it.

License

MIT. See LICENSE. By own2pwn.