CVE-2026-47899: Logseq Path Traversal Vulnerability

CVE-2026-47899 Overview

CVE-2026-47899 is a high-severity vulnerability in the Logseq Electron desktop application. The preload script exposes an API method that lets the renderer process invoke Inter-Process Communication (IPC) handlers without performing path validation. An attacker who achieves JavaScript execution in the renderer, for example through Cross-Site Scripting (XSS) or a malicious plugin, can read, write, or delete arbitrary files on the user's system. Version v0.10.15 was tested and confirmed as vulnerable. The status of other versions remains unconfirmed because no patch has been published. The weakness is tracked as [CWE-749: Exposed Dangerous Method or Function].

Critical Impact
Renderer-side JavaScript execution escalates into arbitrary file read, write, and delete on the host filesystem under the user's privileges.

Affected Products

Logseq desktop application v0.10.15 (Electron build)
Other Logseq desktop versions: status unknown, no vendor patch released
Installations that load third-party Logseq plugins

Discovery Timeline

2026-06-09 - CVE-2026-47899 published to NVD
2026-06-09 - Last updated in NVD database

Technical Details for CVE-2026-47899

Vulnerability Analysis

Logseq is built on Electron, which separates a privileged main process from a sandboxed renderer process. The preload script bridges these contexts and is expected to expose only narrowly scoped, validated functionality to the renderer. In Logseq v0.10.15, the preload script exposes an API surface that allows the renderer to invoke IPC handlers directly. These handlers operate on filesystem paths supplied by the caller but never validate or canonicalize the path argument.

An attacker who controls JavaScript inside the renderer can invoke these handlers with arbitrary absolute or relative paths. The IPC handlers run in the privileged main process and execute filesystem operations using the application's user-level permissions. The result is full read, write, and delete access to any file the user can reach. Renderer code is reachable through note content that triggers XSS, malicious Markdown, or any installed plugin that an attacker can influence.

Root Cause

The root cause is an exposed dangerous function in the Electron preload bridge. The preload script forwards IPC calls without enforcing an allowlist of paths, a workspace root boundary, or input sanitization. Trust is implicitly granted to the renderer, which Electron's threat model treats as untrusted whenever it can render attacker-controlled content.

Attack Vector

Exploitation requires code execution inside the Logseq renderer process. Two realistic paths exist. First, an attacker delivers a Logseq note, imported graph, or clipboard payload that triggers XSS in the renderer. Second, an attacker publishes or tampers with a Logseq plugin that the victim installs. Once renderer JavaScript runs, the attacker calls the exposed preload API and supplies arbitrary filesystem paths. The CVSS vector indicates a local attack vector with no privileges and no user interaction beyond the initial content load, with high impact to confidentiality and integrity.

No verified public exploit code is available. See the CERT PL analysis for technical details on the exposed IPC surface.

Detection Methods for CVE-2026-47899

Indicators of Compromise

Unexpected file modifications under user directories such as ~/.ssh/, ~/.config/, or browser profile folders shortly after Logseq sessions.
Logseq child processes performing filesystem operations on paths outside the configured graph directory.
Newly installed or updated Logseq plugins from untrusted sources preceding suspicious file activity.

Detection Strategies

Monitor the Logseq process tree for file create, write, and delete operations targeting paths outside the user's graph workspace.
Inspect Logseq plugin directories for unsigned or recently modified plugin bundles and audit their package.json permissions.
Apply Endpoint Detection and Response (EDR) behavioral rules that flag Electron renderer processes spawning filesystem syscalls against sensitive directories.

Monitoring Recommendations

Enable filesystem auditing on high-value paths and correlate events with the Logseq process identifier.
Capture and review Logseq plugin installation events and version changes.
Alert on outbound network connections from the Logseq process that follow local file reads, which may indicate exfiltration.

How to Mitigate CVE-2026-47899

Immediate Actions Required

Treat Logseq v0.10.15 as vulnerable and restrict its use on systems holding sensitive files until a vendor fix is available.
Remove untrusted or unverified Logseq plugins and disable plugin auto-update from third-party registries.
Avoid opening Logseq notes, graphs, or Markdown content from untrusted sources.

Patch Information

No vendor patch has been published for CVE-2026-47899 at the time of disclosure. Monitor the Logseq official website and the CERT PL advisory for updates. When a fixed release is published, upgrade immediately and verify that the preload script enforces path validation against the active graph root.

Workarounds

Run Logseq under a dedicated low-privilege user account with no access to sensitive files such as SSH keys, credential stores, or browser profiles.
Use operating system sandboxing such as Firejail on Linux, App Sandbox on macOS, or Windows AppContainer to constrain Logseq's filesystem reach.
Disable the Logseq plugin system entirely in environments where plugins are not required.
Block Logseq from accessing directories outside the designated graph folder using mandatory access control profiles such as AppArmor or SELinux.

bash

# Example AppArmor profile snippet restricting Logseq filesystem access
/usr/bin/logseq {
  # Allow access only to the user's graph directory
  owner @{HOME}/Logseq/** rwk,
  owner @{HOME}/.config/Logseq/** rwk,

  # Deny sensitive paths explicitly
  deny @{HOME}/.ssh/** rwklx,
  deny @{HOME}/.aws/** rwklx,
  deny @{HOME}/.gnupg/** rwklx,
  deny /etc/** wklx,
}