CVE-2025-54135
AI description
CVE-2025-54135 is a security vulnerability in the Cursor AI code editor that could lead to remote code execution. This vulnerability, dubbed "CurXecute" by Aim Labs, arises from the way Cursor interacts with Model Control Protocol (MCP) servers to access external tools. By feeding malicious prompts to the AI agent, an attacker can trigger the execution of attacker-controlled commands. The vulnerability involves a prompt injection that silently rewrites the "~/.cursor/mcp.json" file. This file is used to configure custom MCP servers in Cursor, and the vulnerability allows the execution of any new entry without requiring confirmation. An attacker can inject a malicious command through a single line of prompting, influencing Cursor's actions and gaining remote code execution under the user's privileges.
- Description
- -
Hype score is a measure of social media activity compared against trending CVEs from the past 12 months. Max score 100.
- Hype score
65
01/08/2025 A high-severity flaw (CVE-2025-54135) in Cursor AI Code Editor could enable remote code execution via prompt injection! 🚨 Version 1.3 released to patch this critical vulnerability. Act now to secure your systems! Source: https://t.co/HujC8drGpF
@kernyx64
2 Aug 2025
14 Impressions
0 Retweets
0 Likes
0 Bookmarks
0 Replies
0 Quotes
AIコードエディタ「Cursor」において、リモートコード実行（RCE）が可能となる深刻な脆弱性（CVE-2025-54135、CVSS 8.6）が発見され、2025年7月29日リリースのバージョン1.3で修正された。 この脆弱性「CurXecute」は、
@yousukezan
1 Aug 2025
104717 Impressions
139 Retweets
393 Likes
201 Bookmarks
2 Replies
23 Quotes
El editor de código con IA, Cursor, corrigió una grave falla de ejecución remota de código (RCE), registrada como CVE-2025-54135. Suponía un riesgo crítico para la seguridad de los desarrolladores. El fallo estaba en el protocolo de control del modelo (MCP). Al procesar da
@citarafy
1 Aug 2025
36 Impressions
0 Retweets
1 Like
0 Bookmarks
0 Replies
0 Quotes
A vulnerability named CurXecute (CVE-2025-54135) affects nearly all versions of the AI-powered Cursor IDE, allowing remote code execution via prompt-injection attacks. Exploitation could lead to ransomware and data theft. https://t.co/aXAKQ3NgKQ
@securityRSS
1 Aug 2025
31 Impressions
0 Retweets
0 Likes
0 Bookmarks
0 Replies
0 Quotes
🚨CVE-2025-54135: Cursor AI Code Editor Vulnerability Patched After Prompt Injection Exploit Allowed Remote Command Execution CVSS: 8.6 Video Credit: HiddenLayer Advisory no longer available on GitHub: https://t.co/dNQbdgR9Ff https://t.co/HLZu6k7C7c
@DarkWebInformer
1 Aug 2025
4692 Impressions
5 Retweets
31 Likes
8 Bookmarks
1 Reply
2 Quotes
🚨 A single Slack message could hijack Cursor—AI code editor—with zero clicks. CVE-2025-54135 let attackers run remote code just by posting in a public channel. Cursor auto-executed it. No prompts. No approval. Details here → https://t.co/7dB6UOzzUo
@TheHackersNews
1 Aug 2025
9230 Impressions
27 Retweets
65 Likes
11 Bookmarks
2 Replies
0 Quotes