CVE-2025-6514
CRITICALmcp-remote >=0.0.5 <0.1.16 - OS Command Injection via Authorization Endpoint Response URL
Title source: llmExploitation Summary
EIP tracks 3 public exploits for CVE-2025-6514. PoCs published by Cyberency, darshjme, ChaseHCS.
AI-analyzed exploit summary The repository contains a functional TypeScript-based proxy tool (`mcp-remote`) designed to bridge local MCP clients with remote MCP servers, leveraging OAuth authentication. It includes detailed configuration options for headers, transport strategies, and debugging, indicating a mature and operational proof-of-concept.
Description
mcp-remote is exposed to OS command injection when connecting to untrusted MCP servers due to crafted input from the authorization_endpoint response URL
Exploits (3)
The repository contains a functional TypeScript-based proxy tool (`mcp-remote`) designed to bridge local MCP clients with remote MCP servers, leveraging OAuth authentication. It includes detailed configuration options for headers, transport strategies, and debugging, indicating a mature and operational proof-of-concept.
This repository contains a security audit tool for Model Context Protocol (MCP) servers, focusing on detecting tool poisoning, prompt injection, and input validation issues. It scans for patterns associated with CVE-2025-6514 but does not include functional exploit code.
This repository is a placeholder for CVE-2025-6514, documenting steps to achieve arbitrary code execution in MCP-Remote. It lacks actual exploit code and is a to-do list for future development.
References (3)
Scores
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:H