/

CVE-2026-55450

Severity:

CRITICAL RISK

(9.3)

Unauthenticated File Upload (CWE-400) in Langflow

CVE-2026-55450

Analyzed

Vulnerability type

Category:

Resource Exhaustion / Uncontrolled Upload

CWE:

CWE-200, CWE-306, CWE-400

Impact:

Denial of Service and Information Exposure

Severity:

CRITICAL RISK

(9.3)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

Langflow allows unauthenticated users to upload arbitrary amounts of data to the server, with no size or quota restrictions, leading to potential disk space exhaustion.

The upload response also exposes the absolute filesystem path of the stored file, leaking server layout information that can be combined with other issues under simple network access conditions.

Affected context

This vulnerability occurs in Langflow deployments where the file upload functionality is exposed over the network without authentication or rate/size limiting.

Any reachable instance that accepts uploads from unauthenticated clients and stores them on local disk is affected.

The risk increases when the application runs on shared infrastructure, production servers, or systems with limited disk space or monitoring.

Exploit preconditions

  • Attack surface: Public or internal Langflow HTTP endpoint that accepts file uploads.
  • Required access: Network connectivity to the Langflow server; no authentication or prior knowledge is needed.
  • Trigger point: Sending repeated or very large file upload requests to the affected endpoint and reading the server's response.
  • Why exploitation is possible: The upload logic does not enforce authentication, quota, size limits, or rate limits, and it returns absolute file paths directly in the response.

Why this happens

The underlying issue combines missing access control, resource management, and information disclosure.

From a CWE perspective, it reflects CWE-306: Missing Authentication for Critical Function, where a sensitive file upload feature is exposed without verifying user identity.

It also embodies CWE-400: Uncontrolled Resource Consumption, because incoming data is accepted and persisted without constraints, allowing attackers to exhaust disk resources.

Finally, CWE-200: Exposure of Sensitive Information applies due to the direct inclusion of absolute filesystem paths in responses, revealing internal server structure.

Abstractly, this pattern is an upload endpoint that fully trusts any caller, writes whatever data is provided to disk, and echoes back internal metadata, without defensive boundaries.

Potential impact

  • Attackers can cause disk space exhaustion, leading to application crashes or degraded performance.
  • The Langflow instance may become unavailable, impacting workflows, agents, and dependent services.
  • System instability could affect logging, backups, or other processes sharing the same storage.
  • Exposed absolute paths can assist attackers in mapping the filesystem and chaining further exploits.
  • Operational teams may face unplanned downtime and remediation efforts to recover storage.

General mitigation guidance

Mitigation typically focuses on enforcing access control, constraining resource use, and minimizing information leakage on upload endpoints.

Secure patterns include:

  • Requiring authentication and authorization for file upload operations, aligning access with intended use.
  • Applying strict limits on file size, total data per user, and request rates, with safe defaults.
  • Implementing storage quotas and automatic cleanup policies for temporary or untrusted uploads.
  • Validating and normalizing server responses so they avoid exposing absolute paths or internal layout details.
  • Isolating upload storage (e.g., separate volumes or buckets) and running services with least-privilege filesystem access.
  • Monitoring disk utilization and implementing alerts when storage consumption exceeds expected thresholds.

How teams detect issues like this

High-level detection approaches include:

  • Static analysis or code scanning to identify unauthenticated endpoints that perform file writes or resource-intensive operations.
  • Code review checklists that flag upload handlers lacking size limits, quotas, or rate limiting, and responses that echo internal paths.
  • Logging and monitoring of upload traffic patterns, including unusually large requests or repeated uploads from the same source.
  • Storage monitoring policies that raise alerts on rapid disk usage growth or abnormal file creation rates.
  • Automated security testing (DAST) that probes upload endpoints for missing authentication, unconstrained data acceptance, and path disclosure in responses.

Sources

SourceLink
github.comgithub.com/langflow-ai/langflow/pull/12831
github.comgithub.com/langflow-ai/langflow/security/advisories/GHSA-x223-p2gf-v735

Frequently asked

questions

Any user or script with network access to the Langflow instance can exploit it, because no authentication or special knowledge is required to trigger the upload behavior.

Only basic network connectivity to the vulnerable upload endpoint is needed. The attacker does not need a valid account, credentials, or prior interaction with the system.

The safest approach is to require authentication for file uploads, enforce strict file size and quota limits, and adjust responses so they do not expose absolute paths or other internal metadata.

Design upload and storage features with explicit access control, resource constraints, and privacy-aware response formats, and include checks for these patterns in code review and automated security testing pipelines.

Find out what's already
exploitable in your codebase.