Skip to main content
CVE Vulnerability Database
Vulnerability Database/CVE-2026-14470

CVE-2026-14470: Langflow Path Traversal Vulnerability

CVE-2026-14470 is a path traversal vulnerability in IBM Langflow OSS versions 1.0.0 through 1.10.2 that allows authenticated attackers to access arbitrary files using directory traversal techniques. This article covers technical details, affected versions, security impact, and mitigation strategies.

Published:

CVE-2026-14470 Overview

CVE-2026-14470 is a path traversal vulnerability [CWE-22] affecting IBM Langflow OSS versions 1.0.0 through 1.10.2. An authenticated attacker can send a crafted URL containing ../ sequences to read arbitrary files on the underlying system. The flaw impacts confidentiality of files accessible to the Langflow service process, including configuration files, credentials, and application data. Because Langflow runs on Windows, macOS, and Linux, exposure spans every supported deployment platform. The vulnerability requires low-privilege authentication but no user interaction and is exploitable over the network.

Critical Impact

An authenticated attacker can read arbitrary files on the Langflow host, exposing secrets, source code, and configuration data used by AI workflow pipelines.

Affected Products

  • IBM Langflow OSS versions 1.0.0 through 1.10.2
  • Linux, macOS, and Microsoft Windows deployments of Langflow
  • Downstream applications and pipelines built on affected Langflow releases

Discovery Timeline

  • 2026-09-04 - CVE-2026-14470 published to NVD
  • 2026-09-10 - Last updated in NVD database

Technical Details for CVE-2026-14470

Vulnerability Analysis

Langflow is an open-source visual framework for building applications with large language models. The affected releases expose an HTTP endpoint that accepts a file path parameter without normalizing or constraining it to a permitted base directory. An authenticated attacker can inject ../ traversal sequences into the request path or a path parameter to escape the intended directory and read files elsewhere on the filesystem.

The attacker only needs valid low-privilege credentials, which is common in multi-tenant Langflow deployments where accounts are issued to developers or workflow authors. Successful exploitation exposes sensitive files such as .env configuration files, API keys for LLM providers, database credentials, SSH private keys, and system files like /etc/passwd on Linux hosts. The scope is confidentiality-only: the flaw does not permit modification of files or denial of service.

Root Cause

The root cause is missing input validation on file path parameters accepted by the Langflow web service, tracked as CWE-22 Improper Limitation of a Pathname to a Restricted Directory. The service concatenates attacker-supplied path fragments with a base directory rather than canonicalizing the result and verifying it remains inside the allowed root.

Attack Vector

Exploitation is performed over the network against the Langflow HTTP interface. After authenticating with any valid account, an attacker issues a request that embeds dot-dot-slash sequences in a file-serving or asset-loading endpoint. The server resolves the traversed path and returns the file contents in the HTTP response. Because the request looks like a normal API call from an authenticated user, it blends into standard application traffic.

See the IBM Support Page for the vendor's technical description and fixed version guidance.

Detection Methods for CVE-2026-14470

Indicators of Compromise

  • HTTP request logs containing ../, ..%2f, ..%5c, or double-encoded traversal sequences in URL paths or query parameters against Langflow endpoints
  • Langflow application logs recording file reads outside the expected asset or component directories
  • Outbound access to sensitive files such as /etc/passwd, .env, id_rsa, or Windows configuration files by the Langflow process

Detection Strategies

  • Deploy web application firewall or reverse proxy rules that decode and inspect request paths for traversal patterns before they reach the Langflow service
  • Correlate authenticated Langflow sessions with anomalous request volumes to file-serving endpoints
  • Baseline the set of files Langflow legitimately reads and alert on process-level file opens outside that set

Monitoring Recommendations

  • Forward Langflow access logs, authentication events, and host process telemetry to a centralized analytics platform for retrospective hunting
  • Monitor for successful HTTP 200 responses to requests containing encoded or unencoded traversal sequences
  • Track credential and token access patterns from the Langflow host after any suspected traversal attempts

How to Mitigate CVE-2026-14470

Immediate Actions Required

  • Upgrade Langflow OSS to a version later than 1.10.2 as directed by the IBM Support Page
  • Rotate any secrets, API keys, and credentials stored on or accessible from Langflow hosts running affected versions
  • Audit Langflow user accounts and remove or disable inactive, shared, or unnecessary low-privilege accounts

Patch Information

IBM has published remediation guidance for CVE-2026-14470 on the IBM Support Page. Administrators should apply the fixed release identified in that advisory to all Langflow OSS instances between 1.0.0 and 1.10.2 inclusive.

Workarounds

  • Place Langflow behind a reverse proxy or WAF that normalizes URLs and blocks requests containing ../, ..%2f, and ..%5c sequences
  • Restrict network access to the Langflow interface to trusted administrative networks until patching is complete
  • Run the Langflow process under a dedicated low-privilege OS account with filesystem permissions limited to required directories
bash
# Example NGINX reverse proxy hardening for Langflow
location / {
    if ($request_uri ~* "(\.\./|\.\.%2f|\.\.%5c)") {
        return 400;
    }
    proxy_pass http://langflow_upstream;
    proxy_set_header Host $host;
    proxy_set_header X-Real-IP $remote_addr;
}

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Default Legacy - Prefooter | Experience the World’s Most Advanced Cybersecurity Platform

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.