CVE-2026-104413 in Ghost
Summary
by MITRE • 10/02/2026
Ghost from 5.94.0 before 6.64.0 contains a stored cross-site scripting vulnerability that allows staff users, including Contributors, to host arbitrary HTML by abusing bookmark card image fetching. Attackers can create bookmark cards that store non-image files from external websites as icons or thumbnails to compromise other staff users' admin sessions.
Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.
Analysis
by VulDB Data Team • 10/02/2026
The identified security flaw resides within the Ghost content management system in versions prior to 6.64.0, specifically affecting the functionality introduced since version 5.94.0 regarding bookmark card image fetching mechanisms. This vulnerability is classified as a stored cross-site scripting issue, which represents a severe category of web application insecurity where malicious scripts are permanently embedded within the content of a website or database rather than being executed transiently in a user's browser session alone. The core technical flaw stems from insufficient validation and sanitization processes applied to external resources when they are processed as part of bookmark card creation. When a staff member, including those with lower privilege levels such as Contributors, creates a bookmark card by providing an external URL, the system attempts to fetch metadata and images associated with that link. The vulnerability arises because the application fails to adequately verify whether the fetched resource is actually an image file or if it contains executable script content disguised within headers or embedded data structures intended for icon generation.
This lack of rigorous input validation allows attackers to craft malicious bookmark cards containing non-image files from external websites, which are then stored and rendered as icons or thumbnails on posts or pages visible to other staff users. The operational impact is significant because it enables the persistence of arbitrary HTML and JavaScript code within the application's database. When another authenticated staff member views a page containing this compromised bookmark card, their browser executes the injected script in the context of the Ghost admin interface. This execution occurs with the privileges of the viewing user, effectively allowing an attacker to hijack administrative sessions, steal sensitive cookies or authentication tokens, perform actions on behalf of victims, and potentially escalate privileges if higher-level administrators are targeted. The ability for Contributors, who typically have limited write access but can create public-facing content, to execute such attacks highlights a critical failure in the principle of least privilege enforcement within the application's rendering pipeline.
From an industry standards perspective, this vulnerability aligns with CWE-79, which defines Improper Neutralization of Input During Web Page Generation commonly known as Cross-site Scripting. The specific nature of storing malicious payloads for later execution categorizes it under Stored XSS variants where the payload is saved on target servers and delivered to users via normal means. Furthermore, in terms of tactical mapping within the MITRE ATT&CK framework, this exploit leverages techniques associated with Collection through Browser Session Hijacking or Credential Access by stealing session cookies if the script includes exfiltration logic. It also relates to Defense Evasion as it bypasses standard input filters that might otherwise block direct script injection attempts in text fields, instead exploiting a secondary feature like image fetching for bookmark cards to achieve code execution indirectly.
Mitigation strategies must focus on both immediate patching and long-term architectural improvements. The primary remediation is to upgrade the Ghost installation to version 6.64.0 or later where this specific flaw has been addressed by developers through enhanced input validation and output encoding mechanisms. In addition to upgrading, administrators should implement strict Content Security Policy headers that restrict the execution of inline scripts and limit sources from which content can be loaded, thereby reducing the impact even if a similar vulnerability were present in future versions. Input sanitization libraries should be configured to strictly whitelist allowed HTML tags and attributes for bookmark card rendering, ensuring that only safe image formats are processed by the icon fetching service. Regular security audits focusing on how external resources are handled during content creation workflows will further harden the application against such indirect injection attacks.