CVE-2026-66824
CriticalCVSS 9.2Exploitation Probability (EPSS)
Low risk20th percentile - higher than 20% of all known CVEs
Summary
A stored XSS vulnerability in the capture tree visualization tool. Serialized tree data was embedded in an inline JavaScript block, allowing injection of malicious HTML/JavaScript. An attacker could perform actions in the victim's session context.
Risk Assessment
The risk includes session theft, access to victim data, and modification of application data within user permissions.
Recommendation
Apply the patch that removes JSON data from the HTML document and retrieves it via a dedicated API endpoint.
Other vulnerabilities in Pheditor
See all- CVE-2026-55579Critical
Pheditor ships with a hardcoded default password 'admin' (SHA-512 hash). No mechanism forces a password change on first login. Any deployment using default credentials grants full access to file editor, upload, and terminal, enabling arbitrary file read/write and remote code execution.
- CVE-2026-48030Critical
An OS Command Injection vulnerability in Pheditor allows an authenticated user to execute arbitrary OS commands by injecting shell metacharacters into the 'dir' POST parameter, bypassing the TERMINAL_COMMANDS whitelist.
- CVE-2026-55578High
Pheditor versions 2.0.1-2.0.5 have a vulnerability in the terminal feature with an incomplete character blocklist. Characters |, `, and newline (0x0A) are not blocked, allowing authenticated users to bypass the command allowlist and execute arbitrary OS commands.
- CVE-2026-54540High
Pheditor before version 2.0.5 has a vulnerability in the terminal feature that only checks the start of the command and does not block shell command substitution ($()). Authenticated users can bypass the command allowlist and execute arbitrary OS commands.
Original NVD description (English source)
A stored cross-site scripting vulnerability existed in the capture tree visualization page. The application embedded the serialized capture tree directly into an inline JavaScript block using the Jinja safe filter. Because the tree data can contain values derived from captured and potentially attacker-controlled web content, a specially crafted value could prematurely terminate the surrounding <script> element and inject arbitrary HTML or JavaScript. The malicious code would execute in the browser of a user viewing the affected capture tree. Successful exploitation could allow an attacker to perform actions using the victim’s authenticated session, access information available to the victim, or modify application data within the permissions of the affected user. The patch removes the JSON data from the HTML document and retrieves it through a dedicated API endpoint. The client then processes the response using response.json(), preventing capture data from being interpreted as executable content within the original page’s HTML or JavaScript context.