/

CVE-2026-54309

Severity:

CRITICAL RISK

(10.0)

Missing Auth on MCP Browser HTTP (CWE-306) in n8n

CVE-2026-54309

Analyzed

Vulnerability type

Category:

Missing Authentication

CWE:

CWE-306

Impact:

Unauthorized Browser Control and Account Compromise

Severity:

CRITICAL RISK

(10.0)

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

A missing authentication check in n8n's MCP Browser HTTP transport allows any network-reachable client to initialize sessions and invoke browser-control tools without verification.

When the AI Browser Bridge extension is active, an unauthenticated attacker can remotely drive the user's real browser profile, making this issue critical wherever the MCP endpoint is exposed.

Affected context

This vulnerability arises in n8n deployments that use the @n8n/mcp-browser component in HTTP transport mode (--transport http) without an external authentication or network isolation layer.

It is most relevant when the MCP endpoint is reachable over a local network or the internet and the AI Browser Bridge browser extension is installed and connected.

Under these conditions, the MCP endpoint effectively exposes browser-control capabilities to any party that can send HTTP requests to it.

Exploit preconditions

  • Attack surface: HTTP MCP endpoint created by @n8n/mcp-browser when run with --transport http.
  • Required access: Network reachability to the MCP HTTP port (local network, VPN, or internet, depending on deployment).
  • Trigger point: Sending session initialization and tool invocation requests to the MCP endpoint to establish a browser-control session.
  • Why exploitation is possible: The endpoint does not enforce authentication or session ownership checks, allowing anonymous clients to bind to the user's active browser bridge and issue commands.

Why this happens

The root cause is a missing authentication enforcement on the MCP Browser HTTP transport endpoint, mapped to CWE-306: Missing Authentication for Critical Function.

The service exposes critical browser-control operations—such as navigation, script execution, and access to cookies and storage—without verifying the identity or authorization of the caller.

In abstract terms, a network-facing control channel was designed to trust any client that can connect, instead of treating session initialization and tool invocation as sensitive operations requiring explicit authentication and access control.

Potential impact

  • An attacker may remotely navigate the victim's browser to arbitrary URLs, including phishing or malicious sites.
  • An attacker could read or modify cookies and local storage, potentially compromising authenticated sessions on web applications.
  • An attacker can execute JavaScript in the context of visited sites, which could lead to data exfiltration or account takeover.
  • An attacker may pivot from browser access to broader system compromise by abusing authenticated web applications or sensitive portals.
  • An attacker could perform actions on behalf of the user in SaaS platforms, administration consoles, or developer tooling accessed via the browser.

General mitigation guidance

Mitigation generally focuses on enforcing strict authentication and limiting exposure of remote-control interfaces.

Recommended patterns include:

  • Treating MCP endpoints as privileged control channels and requiring robust authentication and authorization for session establishment and tool invocation.
  • Restricting network access to the MCP Browser HTTP transport via firewall rules, private networks, or localhost-only bindings.
  • Implementing explicit session binding so that only the owning user or process can attach to an active browser bridge.
  • Designing browser automation integrations with least-privilege capabilities, minimizing access to sensitive data (cookies, storage, authenticated sessions) where feasible.
  • Applying secure-by-default configurations so that no unauthenticated HTTP transport is exposed unless consciously enabled with proper safeguards.

How teams detect issues like this

High-level detection approaches may include:

  • Reviewing configuration and deployment manifests to identify @n8n/mcp-browser instances running with --transport http on non-local interfaces.
  • Performing code review to ensure that MCP session initialization and tool invocation endpoints enforce authentication and ownership checks before executing browser-control commands.
  • Using static analysis or automated security testing tools to flag unauthenticated entry points exposing critical operations over HTTP.
  • Monitoring network traffic and logs for unexpected MCP session creation or tool usage originating from unknown clients, particularly on ports associated with the MCP Browser transport.
  • Establishing security policies that prohibit unauthenticated remote-control interfaces and enforce authentication middleware on all externally reachable endpoints.

Sources

SourceLink
github.comgithub.com/n8n-io/n8n/security/advisories/GHSA-qrx8-25qr-5r7v

Frequently asked

questions

Yes. Any attacker who can reach the MCP Browser HTTP endpoint over the network can exploit it, potentially from the internet if the service is exposed or from any site that can cause the client to connect to the endpoint.

The attacker only needs network access to the MCP HTTP transport port. No prior authentication is required to initialize a session or invoke browser-control tools.

Enforce strong authentication and authorization on all MCP endpoints that expose browser-control capabilities, and ensure that only the intended user or trusted components can establish sessions and invoke tools.

Treat remote-control and automation interfaces as high-risk, apply secure-by-default configurations that do not expose unauthenticated HTTP transports, and incorporate authentication, access control, and configuration reviews into development and deployment pipelines.

Find out what's already
exploitable in your codebase.