List of software quality issues with the number of affected components.
category ALL
Policies
Info
Category
Problem
AI assistant extensions are commonly used by software developers to extend the functionality of AI coding assistants, agentic development tools, and other AI-powered developer tools. Extensions are typically distributed through public marketplaces as packages that bundle skill instruction files with supporting resources. Skill instruction files describe specialized capabilities that the AI assistant can invoke, such as guidance for handling specific file formats, performing domain-specific tasks, or interacting with particular services. The AI assistant loads instructions from these files into its context and follows them when relevant user requests or workflow events occur. These instructions are written by extension authors to guide the AI assistant toward the intended capability, or to provide additional context for nominal extension use. However, AI assistant extensions are commonly abused by threat actors to influence the AI assistant into performing arbitrary actions within the development environment. It was detected that the skill instruction files could direct the AI assistant to perform actions that are not typically associated with such tools. Such unusual actions resemble common threat actor tactics and are usually obscured by indirect phrasing, misleading context, or other prompt injection techniques.
If the AI assistant extension intent does not relate to the reported behavior, investigate your development environment for software supply chain compromise.
You should stop using the AI assistant extension until the investigation is completed, or until the issue is risk accepted.
Consider replacing the AI assistant extension with an alternative.
Problem
AI assistant extensions are commonly used by software developers to extend the functionality of AI coding assistants, agentic development tools, and other AI-powered developer tools. Extensions are typically distributed through public marketplaces as packages that bundle skill instruction files with supporting resources. Skill instruction files describe specialized capabilities that the AI assistant can invoke, such as guidance for handling specific file formats, performing domain-specific tasks, or interacting with particular services. The AI assistant loads instructions from these files into its context and follows them when relevant user requests or workflow events occur. These instructions are written by extension authors to guide the AI assistant toward the intended capability, or to provide additional context for nominal extension use. However, AI assistant extensions are commonly abused by threat actors to influence the AI assistant into performing arbitrary actions within the development environment. It was detected that the skill instruction files could direct the AI assistant to execute unusual operating system commands. Unusual commands resemble common threat actor tactics, such as destructive file deletion, elevation of privileges, or tampering with security settings. Unusual skill instruction files often contain indirect phrasing, misleading context, or other prompt injection techniques.
If the AI assistant extension intent does not relate to the reported behavior, investigate your development environment for software supply chain compromise.
You should stop using the AI assistant extension until the investigation is completed, or until the issue is risk accepted.
Consider replacing the AI assistant extension with an alternative.
Problem
AI assistant extensions are commonly used by software developers to extend the functionality of AI coding assistants, agentic development tools, and other AI-powered developer tools. Extensions are typically distributed through public marketplaces as packages that bundle skill instruction files with supporting resources. Skill instruction files describe specialized capabilities that the AI assistant can invoke, such as guidance for handling specific file formats, performing domain-specific tasks, or interacting with particular services. The AI assistant loads instructions from these files into its context and follows them when relevant user requests or workflow events occur. These instructions are written by extension authors to guide the AI assistant toward the intended capability, or to provide additional context for nominal extension use. It is unusual for certain types of AI assistant extensions to direct the AI assistant to download additional content from a remote server. Attackers commonly abuse AI assistant extensions to fetch malicious payloads from public code repositories, file sharing websites, or their own infrastructure. Remotely hosted content is not immutable, allowing the attackers to change the type of malware they deploy at any time.
If the AI assistant extension intent does not relate to the reported behavior, investigate your development environment for software supply chain compromise.
You should stop using the AI assistant extension until the investigation is completed, or until the issue is risk accepted.
Consider replacing the AI assistant extension with an alternative.
Problem
Operating systems execute application code in multiple privilege access levels. Separation of privileges is designed to protect the stability and integrity of the operating system by shielding it from issues that the user run applications may cause. However, some users may need to interact with higher privilege parts of the operating system to accomplish specific tasks. For this purpose, operating systems provide facilities that users may leverage to temporarily elevate their running privileges. Users with higher privileges can run any application with the same privilege level as their own. Attackers often try to trick privileged users into running malicious code, enabling them to infect the operating system. While the presence of code that elevates user privileges does not necessarily imply malicious intent, all of its uses in a software package should be documented and approved. Only select applications should consider using functions that can elevate user privileges. One example of acceptable use for such functions is allowing the users to install software packages and updates.
Prevalence in PyPI community
2 packages
found in
Top 100
35 packages
found in
Top 1k
215 packages
found in
Top 10k
7.5k packages
in community
Next steps
Investigate reported detections as indicators of software tampering.
Consult Mitre ATT&CK documentation: T1548 - Abuse Elevation Control Mechanism.
Consider rewriting the flagged code without using the marked behaviors.
Problem
Software components are typically distributed in standardized packaging formats. Software packages are built from instructions written within package manifests that act as blueprints for package assembly. A package manifest declares the most important software properties, such as the package name, its authors and license, external dependencies, and various actions that may occur during the package lifecycle. Package managers are specialized tools used by developers to manage software components. Package managers read these manifests to deploy software components along with their dependencies. The premise of the package management system is that software components are isolated from each other, and that they rely on the package manager to provide their dependencies. However, it was detected that a software component includes code capable of accessing or modifying directories that are maintained exclusively by package managers. This is unusual as it resembles tactics used by threat actors that tamper with installed software components and inject malicious code into trusted execution paths.
Prevalence in PyPI community
8 packages
found in
Top 100
50 packages
found in
Top 1k
303 packages
found in
Top 10k
11.8k packages
in community
Next steps
Investigate reported detections as indicators of software tampering.
Consider rewriting the flagged code without using the marked behaviors.
Problem
Software components sometimes need to interact with sensitive parts of the runtime environment, often requiring privileged access to accomplish a task. Build pipeline configuration includes high-value information such as CI/CD platform secrets and tokens, pipeline definition files and runner configuration, build-time environment variables containing signing keys or deploy credentials, and artifact registry and package publishing credentials. That information is used by continuous integration and delivery systems. Attackers often abuse access to build pipeline configuration to harvest secrets from build environments, which can later be exfiltrated to attacker-controlled infrastructure and used to tamper with software releases or publish malicious artifacts under trusted identities. While the presence of code that accesses build pipeline configuration does not necessarily imply malicious intent, all of its uses in a software package should be documented and approved. Only select software components should consider using functions that read build pipeline configuration. One example of acceptable use for such functions is release tooling that reads publishing credentials to push artifacts.
Prevalence in PyPI community
0 packages
found in
Top 100
16 packages
found in
Top 1k
58 packages
found in
Top 10k
2.44k packages
in community
Next steps
Investigate reported detections as indicators of software tampering.
Consider rewriting the flagged code without using the marked behaviors.
Problem
Uniform Resource Locators (URLs) are structured addresses that point to locations and assets on the internet. URLs allow software developers to build complex applications that exchange data with servers that can be hosted in multiple geographical regions. URLs can commonly be found embedded in documentation, configuration files, source code and compiled binaries. One or more embedded URLs were discovered to link to raw files hosted on GitHub. Attackers often abuse popular web services to host malicious payloads. Since code-sharing service URLs are typically allowed by security solutions, using them for payload delivery increases the odds that the malicious code will reach the user. While the presence of code-sharing service locations does not imply malicious intent, all of their uses in a software package should be documented and approved. An increasing number of software supply chain attacks in the open source space leverage the GitHub service to deliver malicious payloads.
Prevalence in PyPI community
37 packages
found in
Top 100
233 packages
found in
Top 1k
1795 packages
found in
Top 10k
92.7k packages
in community
Next steps
Investigate reported detections.
If the software should not include these network references, investigate your build and release environment for software supply chain compromise.
You should delay the software release until the investigation is completed, or until the issue is risk accepted.
Consider an alternative delivery mechanism for software packages.
Problem
Applications communicate with web services by exchanging HTTP requests. During software development, externally hosted services are used by developers to debug software quality issues relating to exchanging HTTP requests. Attackers commonly abuse tools designed for HTTP request inspection to monitor network traffic and extract sensitive information from the HTTP traffic. While the presence of domains related to HTTP inspection does not imply malicious intent, all of their uses in a software package should be documented and approved. Attackers might have purposely injected security testing tools in the software package to monitor the network traffic of the infected computer system. It is also possible that the software package has mistakenly included a part of its testing infrastructure during packaging.
Prevalence in PyPI community
11 packages
found in
Top 100
46 packages
found in
Top 1k
254 packages
found in
Top 10k
14.37k packages
in community
Next steps
Investigate reported detections.
If the software should not include these network references, investigate your build and release environment for software supply chain compromise.
You should delay the software release until the investigation is completed, or until the issue is risk accepted.
Consider removing all references to flagged network locations.
Problem
Software developers use programming and design knowledge to build reusable software components. Software components are the basic building blocks for modern applications. Software consumed by an enterprise consists of hundreds, and sometimes even thousands of open source components. Software developers publish components they have authored to public repositories. While a new software project is a welcome addition to the open source community, it is not always prudent to indiscriminately use the latest components when building a commercial application. Irrespective of the software quality, the danger of being the first to try out a new project lies in the fact that the software component may contain novel, currently undetected malicious code. Therefore, it is prudent to review software component behaviors and even try out software component in a sandbox, an environment meant for testing untrusted code.
Prevalence in PyPI community
2 packages
found in
Top 100
22 packages
found in
Top 1k
623 packages
found in
Top 10k
614.08k packages
in community
Next steps
Check the software component behaviors for anomalies.
Consider exploratory software component testing within a sandbox environment.
Consider replacing the software component with a more widely used alternative.
Avoid using this software package until it is vetted as safe.
Problem
Software components contain executable code that performs actions implemented during its development. These actions are called behaviors. In the analysis report, behaviors are presented as human-readable descriptions that best match the underlying code intent. While most behaviors are benign, some are commonly abused by malicious software with the intent to cause harm. When a software package shares behavior traits with malicious software, it may become flagged by security solutions. Any detection from security solutions can cause friction for the end-users during software deployment. While the behavior is likely intended by the developer, there is a small chance this detection is true positive, and an early indication of a software supply chain attack.
Prevalence in PyPI community
20 packages
found in
Top 100
103 packages
found in
Top 1k
958 packages
found in
Top 10k
55.62k packages
in community
Next steps
Investigate reported detections.
If the software intent does not relate to the reported behavior, investigate your build and release environment for software supply chain compromise.
You should delay the software release until the investigation is completed, or until the issue is risk accepted.
Consider rewriting the flagged code without using the marked behaviors.