CVE-2025-3248: A critical flaw has been discovered in Langflow that allows malicious actors to execute arbitrary Python code on the target system. This can lead to full remote code execution without authentication, potentially giving attackers control over the server.# CVE-2025-3248 – Execute arbitrary Python code on vulnerable Langflow server.
**Discovered by:** Naveen Sunkavally, Horizon3.ai
**Status:** Public Disclosure
**Date Reported:** 2025-04-07
**CVE ID:** CVE-2025-3248
---
## 🧾 Screenshot
---
## 🧩 Summary
A critical flaw has been discovered in Langflow that allows malicious actors to execute arbitrary Python code on the target system. This can lead to full remote code execution without authentication, potentially giving attackers control over the server. The issue stems from unsafe handling of code input, where user-supplied Python is executed without proper validation or sandboxing.
---
## 📈 Impact
- 🔧 **Remote Code Execution** – Run arbitrary Python code or system commands.
- 🔐 **Full Server Compromise** – Gain control of the system if Langflow runs with elevated privileges.
- 📂 **Data Exposure** – Access sensitive environment variables, credentials, and internal files.
- 🕳️ **Persistence** – Install reverse shells or backdoors for long-term access.
**CVSS Estimate:** Critical (Base Score ~9.8)
---
## 🧾 Affected version
- affected from 0 through 1.2.0
---
## Installation
You should first install the vulnerable lab, theres a guide on the [Docker](https://github.com/vigilante-1337/CVE-2025-3248/blob/main/Docker/setup) folder
```bash
~$ git clone https://github.com/vigilante-1337/CVE-2025-3248
~$ cd CVE-2025-3248
~$ chmod +x exploit
~$ ./exploit -h
```
---
## 🧾 References
- [CVE-2025-3248 on MITRE (pending)](https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2025-3248)
---
_Disclosure coordinated with respect to responsible disclosure guidelines. This document is released for educational and defensive purposes._
登录后查看神龙缓存的 POC 文件快照
登录查看