Brotli vulnerability threatens some WordPress servers
The Brotli vulnerability tracked as CVE-2026-111497 can allow code execution through some WordPress sites, even though the flaw is not in WordPress Core, a plugin or a theme. Google published the vulnerability on October 22, 2026, one day after the Brotli project issued corrected versions. It has a Common Vulnerability Scoring System score of 9.2 out of 10, indicating critical severity. The unofficial names HTTP2Shell, Node2Shell and Brotli2Shell have circulated, although the flaw does not involve the HTTP/2 protocol.
Brotli is a format and software library that compresses web content so it takes less data to transfer. CVE-2026-111497 is an out-of-bounds write, meaning malicious compressed data can make the decoder write beyond its allocated area of memory. That may crash a process, expose information held in memory or produce unauthenticated remote code execution, which lets an attacker run commands without first signing in. Exploitation requires attacker-controlled compressed data to reach an affected decoder in a process exposed through the network. Finding Brotli installed on a server therefore does not, by itself, prove that a website is remotely exploitable.
The affected upstream releases are Brotli 1.0.3 through 1.0.9, fixed in 1.0.10; 1.1.0, fixed in 1.1.1; and 1.2.0, fixed in 1.2.1. These affected versions apply across platforms. Operating-system and hosting vendors may backport a correction, which means placing the security fix into an older package without changing its main version number. Debian 13, for example, fixed the issue in libbrotli1 1.1.0-2+deb13u1, although that package still starts with 1.1.0. Administrators must compare the complete package revision and consult the vendor notice rather than checking only the first three version numbers.
The public WordPress proof of concept uses XML-RPC, a WordPress interface through which other systems can call site functions. An unauthenticated request invokes pingback.ping at xmlrpc.php and provides an attacker-controlled source address. WordPress retrieves that address to verify that it links to a real post. The remote server answers with a specially constructed Brotli-compressed response, which is processed by the vulnerable decoder. If the payload matches the server software and memory layout, it can run code with the permissions assigned to the website's PHP process.
Several conditions must align. xmlrpc.php must be reachable, pingback.ping must be registered, the selected post must accept pings, outbound access must be permitted, and the site's HTTP client must use the affected decoder. The currently published exploit must also closely match the PHP build, runtime and memory arrangement. XML-RPC is the trigger in this demonstration, not a requirement of the underlying Brotli flaw. Other software can expose the decoder, while plugins or themes could create additional WordPress routes. In its scoped review of stock WordPress Core, Wordfence found pingback.ping to be the clear route that lets an unauthenticated attacker provide a public address; other examined fetching features required authentication, extra permissions or previous control of content.
Wordfence reproduced code execution in isolated containers matching the proof of concept's intended environment: WordPress 7.1.3 on Debian 13.7, Apache 2.4.69, PHP 8.3.36 loaded into Apache as mod_php, and the vulnerable Debian libbrotli1 1.1.0-2+b7 package. Standard operating-system memory randomization was active. After only Brotli was updated to Debian's corrected package, the same sequence no longer ran code and the server remained healthy.
The identical payload did not execute against PHP 8.3.35. In one vulnerable WordPress environment, it crashed the PHP worker instead. Wordfence said this demonstrates that a PHP version alone cannot predict whether an attempt will execute code, cause a crash or fail. Attackers can often identify WordPress, PHP, the web server, XML-RPC availability and some outbound-request behavior remotely, potentially helping them select payloads for common server images. Ordinary web requests do not normally disclose the exact binary build and memory layout required to adapt this public payload to any arbitrary server, and Wordfence did not demonstrate such remote adaptation. Its assessment is that the current exploit is not universal, although targeted attacks or reuse against standardized server images remain plausible.
On October 27, 2026, the Cybersecurity and Infrastructure Security Agency added CVE-2026-111497 to its Known Exploited Vulnerabilities catalog. The public entry does not identify the affected application, victim, route or outcome, and does not say WordPress was involved. ELLIO captured an attempt aimed at an official Node.js build, but it failed to produce a shell during 50 replays against a matching vulnerable target. Previdian continued to see attempts, including on November 2, without telemetry showing whether targets were vulnerable or attacks succeeded. Check Point reported automated scanning and suspicious payloads, but no confirmed compromise. Imperva and ELLIO separately achieved code execution against vulnerable Node.js builds in controlled laboratories. Those tests establish that exploitation is possible beyond WordPress, not that production systems were compromised.
As of November 2, 2026, Wordfence found no independently reviewable public account of a successful production compromise or compromise through the WordPress XML-RPC route. That absence does not establish that no exploitation occurred, particularly because CISA has not published incident details. Confirmed laboratory code execution, continuing attempts and the catalog listing still make server-level patching urgent.
The WordPress version under Tools > Site Health cannot determine exposure because Brotli may come from the operating system, hosting platform, container image, web server or PHP stack. Managed-hosting customers should ask whether every Brotli decoder used by the site is corrected for CVE-2026-111497 and whether all affected services were restarted or redeployed. Self-managed administrators should install their platform vendor's update and restart Apache, PHP-FPM, which is a service that runs PHP separately from the web server, and any other process that loaded the old library.
A container image is the packaged filesystem and software used to create containers. Container users must rebuild and redeploy from an updated base image because restarting an existing container does not replace vulnerable files. CloudLinux and similar platforms may hold separate system, web-server or bundled PHP copies of Brotli, so correcting one copy may leave others exposed. Agencies and providers should review production, staging and legacy environments rather than assuming one package check covers every PHP pool, container or server image.