September 2025 Summaries

Software supply chain attacks have significantly increased, particularly through malware in open source dependencies, posing a substantial threat to application security and engineering teams. Since 2023, there has been a 1,300% rise in malware campaigns targeting popular registries like npm and PyPI, with 18,000 malicious packages identified in the first quarter of 2025. This surge has led to 80% of DevOps teams experiencing delays in continuous integration pipelines due to critical security issues, as popular npm packages often have over 100,000 downstream dependencies. Endor Labs offers proactive solutions by detecting and blocking malware at the source, enforcing development guardrails, and reducing incident response workloads. Their approach includes blocking known malicious packages, identifying suspicious behaviors such as typosquatting and dependency confusion, and relying on expert verification to escalate and notify teams of detected threats. Additional measures involve setting guardrails to prevent unpinned dependencies, enforcing cooldown periods for adopting new versions, and assessing every open-source package against stringent health checks to flag weak security practices and risky activities.

The recent surge in npm attacks highlights the need for enterprises to rethink their approach to open-source software (OSS) security by implementing Zero Trust principles, which advocate for never trusting and always verifying. Modern applications heavily rely on OSS, with an average of 80% of codebases sourced from external packages, often introduced by AI coding assistants, creating new attack vectors. Current application security practices assume OSS is safe if it passes vulnerability scans, but Zero Trust suggests every dependency should be continuously verified for risks such as new vulnerabilities or maintainer compromise. This approach involves adapting principles like default deny, continuous verification, least privilege, assume compromise, and ensuring visibility and provenance of OSS packages. Practical measures, such as pinning dependencies, delaying adoption of new packages, and using risk scoring tools, are essential to enforce these principles. Endor Labs provides tools and automated systems to implement Zero Trust for OSS, offering high-fidelity code scanning, risk intelligence, flexible policy enforcement, and evidence-based remediation to protect against supply chain threats while maintaining developer productivity. The adoption of Zero Trust for OSS is becoming a critical baseline for application security, as highlighted by initiatives like the OWASP Top 10 Risks for Open Source Software.

Recent malware campaigns targeting npm packages have highlighted the need for proactive defenses in development environments. These attacks, such as the Shai-Hulud worm and s1ngularity NX, typically target developers to steal credentials and infiltrate larger systems, although they are often short-lived, with malicious packages being discovered and revoked quickly. To mitigate these threats, organizations are encouraged to implement strategies like dependency pinning and lockfiles to ensure deterministic builds and prevent unexpected updates. Additionally, instituting a cooldown period of 24-48 hours before adopting new package versions can allow time for the community to detect and flag malicious releases. Endor Labs offers solutions for enforcing these proactive measures through custom policies using Open Policy Agent and Rego, allowing teams to define specific rules and conditions to safeguard their software supply chain against malware.

Recent months have seen a surge in malware attacks, particularly targeting the npm ecosystem, highlighting the need for heightened vigilance and proactive measures in application security. Unlike CVEs, which are unintentional software flaws documented for remediation, malware is deliberately introduced by attackers with the intent to compromise systems, often targeting developers and their environments to exfiltrate sensitive data. Key differences between malware and CVEs include the short lifespan of malicious packages, as they are quickly removed once detected, and their focus on compromising systems rather than exploiting existing vulnerabilities. The npm ecosystem is especially vulnerable due to JavaScript's widespread use and certain insecure defaults, making it an attractive target for attackers. A series of high-profile attacks throughout the summer of 2025 demonstrated the ease with which attackers could exploit these vulnerabilities, leading to significant impacts on popular packages. To mitigate these risks, organizations are encouraged to adopt proactive strategies like endpoint detection, integrity checks, and delay in package adoption, along with fostering secure practices among developers to reduce the likelihood and impact of such attacks.

In a conversation between Travis McPeak of Cursor (Anysphere) and Jenn Gile from Endor Labs, the discussion revolves around the current state and challenges of application security, particularly concerning malicious packages and industry trends. McPeak reflects on the unchanged nature of security concerns over the past decade, highlighting how awareness has grown due to recent high-profile vulnerabilities. He emphasizes the importance of integrating effective security measures without burdening engineers with constant updates, advocating for the use of AI to enhance security practices by providing quick insights and reducing noise in vulnerability alerts. McPeak explains Cursor's approach to security, which includes ensuring the safety of its AI-driven tools and retaining customer trust by managing sensitive data responsibly. The conversation also touches on the role of software composition analysis tools in prioritizing and addressing vulnerabilities, with McPeak expressing satisfaction with Endor Labs' ability to significantly reduce irrelevant alerts, thereby streamlining security efforts. The discussion concludes with McPeak's preference for maintaining a seamless partnership with engineering teams by solving security issues systemically rather than instituting a security champions program.

NPM has become a frequent target for attackers due to its extensive ecosystem, where compromised accounts can rapidly spread malicious packages across numerous applications and organizations. Recent incidents, such as the "Shai-Hulud" campaign, have highlighted vulnerabilities in the NPM ecosystem, where attackers use stolen credentials to infiltrate accounts and propagate malware through automatic scripts. The high number of transitive dependencies in NPM increases the risk, as a single compromised package can affect many others, creating a large attack surface. Security experts recommend reducing risks by hardening build pipelines, using integrity checks, employing lockfiles, and integrating malware detection in CI/CD processes. Developers are advised to secure their accounts with measures like two-factor authentication, carefully manage dependencies, and adopt safe coding practices. As supply chain attacks persist, the emphasis is on mitigating rather than preventing risks through layered defense strategies and enhancing security awareness across teams.

Endor Labs, based in Palo Alto, is experiencing significant growth in the application security sector, driven by the rising demand for secure development amidst increasing AI adoption. With a 225% year-over-year revenue growth, the company is now protecting 7.4 million applications for major clients like OpenAI, Robinhood, and Dropbox, showcasing its platform's ability to offer precise analysis and evidence-based remediation without hindering engineering productivity. As AI accelerates product cycles and introduces new security vulnerabilities, Endor Labs' comprehensive approach combines deep code visibility, proprietary vulnerability data, and seamless integrations into AI-native Software Development Life Cycles (SDLCs) to reduce noise by 92%, helping engineers address vulnerabilities efficiently. Recent customer acquisitions, including Atlassian, highlight the trust in Endor Labs' proactive security measures tailored to meet the challenges posed by AI-generated code. The company continues to expand its capabilities, launching new tools such as improved Software Composition Analysis (SCA) for C/C++ and an AI Security Code Review feature that anticipates potential design flaws. With $163 million raised to date, Endor Labs is poised to address ongoing application security challenges and set the pace for secure software development in the AI era.

A recent supply chain attack affecting npm packages has been identified, with over 582 compromised package versions from 194 distinct packages, including popular ones like @ctrl/tinycolor. The attack, which began on September 15, involves malicious code that spreads like a virus by using the npm credentials of compromised developers to infect other packages. This code exfiltrates secrets from infected systems and GitHub repositories and replicates itself across npm packages. While some malicious packages remain available, organizations are urged to take immediate preventative actions, such as stopping npm usage in CI/CD pipelines and reviewing package-lock.json files for references to infected versions. Mitigation strategies include using lockfiles pinned to known-good versions, cleaning caches, and implementing cooldown options to minimize the risk of downloading compromised packages. The attack's full impact remains unknown, and efforts to identify and remove infected versions continue.

An open source maintainer's npm account, known as Qix, was compromised through a phishing attack, allowing attackers to inject malicious code into 25 widely used npm packages, including "chalk" and "debug." These packages, which collectively receive hundreds of millions of downloads weekly, are embedded in many popular frameworks and tools, thereby expanding the impact of the attack. The injected malware attempts to steal sensitive data and can further compromise developer systems and CI/CD pipelines. The attack specifically targets cryptocurrency transactions, silently hijacking and redirecting funds to attacker-controlled accounts. This incident underscores the vulnerability of the open source ecosystem, where the compromise of a single maintainer can have widespread ramifications. Organizations are advised to downgrade to safe versions of the affected packages, audit their projects, and enhance account security measures such as enabling two-factor authentication and using scoped access tokens.