CVE-2026-4810 Scanner
CVE-2026-4810 Scanner - Code Injection vulnerability in Google ADK-Python
Short Info
Level
Single Scan
Single Scan
Can be used by
Asset Owner
Estimated Time
10 seconds
Time Interval
22 days 11 hours
Scan only one
Domain, Subdomain, IPv4
Toolbox
Google ADK-Python is utilized by developers and organizations who require an Agent Development Kit for building and deploying voice agents and applications with Google's Cloud services. It is employed in a variety of environments including Python (OSS), Cloud Run, and Google Kubernetes Engine (GKE), enabling developers to create and manage agents efficiently. The software is crucial for integrating conversational interfaces in applications, making it widely adopted in the technology sector. Organizations using the Google ADK-Python rely on its robust features to streamline their development processes and deploy scalable solutions in production and local environments. The kit is intended to support dynamic and interactive experiences, which are essential for modern application development, enabling seamless connection with Google's extensive service offerings.
The Code Injection vulnerability in Google ADK-Python allows for unauthorized execution of arbitrary code on the server. This security flaw is critical as it requires no authentication, opening up systems to remote attackers who can exploit this weakness. Code Injection vulnerabilities like this can result in full system control by an attacker, posing severe risks to affected environments. Such vulnerabilities are primarily due to insufficient input validation and improper authentication mechanisms that fail to restrict unauthorized access. This particular vulnerability impacts multiple versions of Google ADK-Python, making it essential for affected systems to upgrade immediately. Proper understanding of the vulnerability can aid organizations in preventing unauthorized code execution on their servers.
Technical details of the vulnerability indicate that the endpoint '/builder/save' is susceptible to exploitation through improper input handling, particularly with the 'agent.py' file. Attackers can make HTTP POST requests to this endpoint using multipart form-data to deliver the malicious payload without needing authentication. The vulnerability lies in the absence of effective ID validation and control over the input data, allowing malicious actors to force the system into executing their chosen code. Furthermore, this vulnerability is present in the '/list-apps?detailed=true' endpoint, susceptible to unauthorized detailed enumeration, further exposing sensitive data. The combination of inappropriate request handling and lack of authentication protocols makes these endpoints critical points of vulnerability.
Possible effects of exploiting this vulnerability include unauthorized control by attackers over the server, potentially leading to data theft, system corruption, and denial of service. Code Injection can allow malicious users to deploy their code, leveraging the system's resources for unintended purposes or accessing and altering sensitive information. If exploited, systems might be manipulated to serve as a launchpad for further attacks or be incapacitated by a denial-of-service attack. Immediate remediation is essential as the full compromise of affected systems can lead to irreversible damage and significant data breaches.
REFERENCES