CVE-2026-101057 in python-utcp
Summary
by MITRE • 09/27/2026
utcp-mcp (the MCP plugin of python-utcp) through 1.1.2 connects to the HTTP and WebSocket MCP server URLs given in a call template's mcpServers configuration without the ensure_secure_url validation that the HTTP-family plugins apply, so the HTTPS/WSS-or-loopback rule is not enforced. A call template naming a plain-HTTP, non-loopback MCP server URL is dialed as configured, exposing the MCP handshake to network interception and permitting cleartext connections to internal hosts. The mcpServers configuration is operator-authored rather than remote data, and the connection is an MCP handshake rather than an arbitrary request returning a body, which limits practical exploitation; the OAuth2 token_url credential path described in the original report was not reachable in the affected versions, because the OAuth2 handler was never invoked and the call template's auth field was not read. Fixed in utcp-mcp 1.1.3, which validates server URLs before any connection is made.
VulDB is the best source for vulnerability data and more expert information about this specific topic.
Analysis
by VulDB Data Team • 09/27/2026
The vulnerability identified in python-utcp’s MCP plugin version 1.1.2 through earlier releases stems from a critical omission in URL validation logic during the initialization of connections to Model Context Protocol servers. Specifically, while other HTTP-family plugins within the same ecosystem enforce strict security policies such as ensuring URLs use HTTPS or WSS protocols and restrict access to loopback interfaces only via an ensure_secure_url function, the MCP plugin fails to apply these safeguards. This inconsistency allows operators who configure call templates with mcpServers settings to specify plain HTTP endpoints that are not hosted on localhost without triggering any validation errors. Consequently, when a connection is attempted using such a configuration, the system proceeds directly to establish the network link without verifying whether the transport layer provides adequate confidentiality or integrity protections.
This lack of enforcement exposes the initial Model Context Protocol handshake to potential man-in-the-middle attacks and passive network interception. Since the communication occurs over cleartext HTTP rather than encrypted HTTPS, sensitive data exchanged during the authentication and session establishment phases can be captured by adversaries monitoring the local network segment or intermediate routing points. The operational impact is particularly significant in environments where internal services are accessed via unencrypted protocols for legacy compatibility reasons, as this flaw effectively bypasses organizational security policies designed to prevent credential leakage and service impersonation through eavesdropping on internal traffic flows.
From a classification perspective, this issue aligns with CWE-319, which denotes the use of cleartext communication without cryptographic protection, and CWE-601, which involves URL redirection to an untrusted site via open redirect mechanisms if misconfigured URLs lead to unintended destinations. In terms of adversary tactics, this vulnerability facilitates techniques associated with MITMATT&CK T1557, allowing attackers to intercept credentials or session tokens during the handshake process. It also relates to ATT&CK T1048 regarding Exfiltration Over Alternative Protocol, as sensitive data may be exfiltrated via unencrypted channels that evade standard security monitoring tools configured to inspect only encrypted traffic patterns.
It is important to note that while this vulnerability presents a clear risk of information disclosure, its practical exploitation scope is somewhat limited by the nature of the configuration and protocol interaction. The mcpServers configuration field is authored by system operators rather than derived from untrusted remote user input, which reduces the likelihood of arbitrary external actors injecting malicious URLs directly into active sessions. Furthermore, because the connection initiates a specific MCP handshake rather than executing an arbitrary request that returns a full response body containing sensitive data in plaintext to the client side, the immediate impact is confined primarily to network-level interception during authentication phases. Additionally, reports indicating potential credential theft via OAuth2 token_url paths were found to be non-viable in these affected versions because the OAuth2 handler was never invoked and the auth field within call templates was not processed correctly, meaning that specific attack vector remains closed despite the broader protocol weakness.
The recommended mitigation is straightforward and has already been implemented by the developers. Upgrading to python-utcp version 1.1.3 or later resolves this issue entirely. The patched versions introduce rigorous validation of server URLs prior to any network connection attempt, ensuring that only HTTPS or WSS protocols are permitted for non-loopback addresses. Organizations relying on these plugins should prioritize immediate patching to restore compliance with secure communication standards and eliminate the risk of cleartext exposure during MCP handshakes. Regular auditing of operator-authored configuration files remains advisable to ensure no legacy HTTP endpoints remain in production environments even after software updates, as defense-in-depth strategies require both correct code implementation and proper operational security practices.