Overview
In the world of programming, code editors play a crucial role. One such editor, Cursor, which is built specifically for AI programming, has been identified with a severe vulnerability (CVE-2025-61591) in versions 1.7 and below. This vulnerability, if exploited, could lead to system compromise or data leakage, posing significant security threats to organizations and individuals using the affected versions of Cursor. The exploit works when MCP uses OAuth authentication with an untrusted MCP server, leading to potential remote code execution and command injection.
Vulnerability Summary
CVE ID: CVE-2025-61591
Severity: High (8.8)
Attack Vector: Network
Privileges Required: Low
User Interaction: Required
Impact: Potential system compromise or data leakage
Affected Products
Escape the Surveillance Era
Most apps won’t tell you the truth.
They’re part of the problem.
Phone numbers. Emails. Profiles. Logs.
It’s all fuel for surveillance.
Ameeba Chat gives you a way out.
- • No phone number
- • No email
- • No personal info
- • Anonymous aliases
- • End-to-end encrypted
Chat without a trace.
Product | Affected Versions
Cursor | 1.7 and below
How the Exploit Works
The exploit works when the Cursor code editor interacts with an untrusted MCP server using OAuth authentication. During this interaction, an attacker can impersonate a malicious MCP server and return crafted, maliciously injected commands. These commands could allow arbitrary code execution on the host by the agent, leading to a system compromise. The attacker could then run arbitrary commands with full user privileges, potentially leading to data leakage or further system compromise.
Conceptual Example Code
Below is a conceptual example of how this vulnerability might be exploited:
POST /mcp/oauth/token HTTP/1.1
Host: untrusted-mcp-server.com
Content-Type: application/json
{
"grant_type": "authorization_code",
"code": "AUTH_CODE",
"redirect_uri": "http://localhost:8080/callback",
"client_id": "CLIENT_ID",
"client_secret": "CLIENT_SECRET",
"commands": "{malicious_commands}"
}In this example, the attacker impersonates the MCP server and sends a response containing malicious commands in the `commands` field. These commands could be crafted to exploit the command injection vulnerability in the Cursor code editor, leading to arbitrary code execution.
Mitigation Guidance
While the issue has no fixed release version yet, a patch (2025.09.17-25b418f) has been provided by the vendor. Users are strongly advised to apply this patch immediately to mitigate the risk. In case the patch cannot be applied immediately, using a Web Application Firewall (WAF) or an Intrusion Detection System (IDS) can serve as temporary mitigation. These systems can help detect and block malicious traffic, thereby reducing the risk of exploitation. Users should also ensure only trusted MCP servers are used for OAuth authentication to avoid potential exploitation.
