Justin Cappos

The topic of Cappos' Ph.D. dissertation at the University of Arizona was the Stork Project,[10] a software package manager he built with John H. Hartman, a professor in the Department of Computer Science. Stork is still used today in some applications, but, more importantly, the project called attention to the need for improved security for software update processes, a research area Cappos has continued to pursue.

While a post-doctoral researcher at the University of Washington in 2009, Cappos also developed a peer-to-peer computing platform called Seattle,[11][12] which allows device-to-device connectivity in a decentralized network. Seattle is currently used by thousands of developers, who can access, download, and use the program on any type of smart device. In addition, spin-off technologies, such as Sensibility Testbed,[13] have extended the use of Seattle's security and enforced privacy protection strategies, allowing researchers to collect data from sensors at no risk to the privacy of the device owner.

In 2010, Cappos developed The Update Framework (TUF),[14][15] a flexible software framework that builds system resilience against key compromises and other attacks that can threaten the integrity of a repository.[16][17] TUF was designed for easy integration into the native programming languages of existing update systems, and since its inception, it has been adopted or is in the process of being integrated by a number of high-profile open-source projects. One of the more significant earlier adoptions was Docker Content Trust.[18][19] an implementation of the Notary project from Docker that deploys Linux containers.[20] Notary, which is built on TUF, can certify the validity of the sources of Docker images.[21] In October 2017, Notary and TUF were both adopted as hosted projects by the Linux Foundation as part of its Cloud Native Computing Foundation.[22] In December 2019, TUF became the first specification and first security-focused project to graduate from CNCF.[23] TUF has also been standardized in Python,[24][25] and been independently implemented in the Go language by Flynn, an open-source platform as a service (PaaS) for running applications in production.[26][27][28] To date, the list of tech companies and organizations using TUF include Foundries.io[29],IBM,[30] VMware,[31] Digital Ocean,[32] Microsoft,[33] Google,[34] Amazon,[35] Leap,[36] Kolide,[37] Docker,[38] and Cloudflare.[39]

Another significant compromise-resilient software update framework by Cappos is the 2017 launch of a TUF-adapted technology called Uptane.[40][41] Uptane is designed to secure software updates for automobiles, particularly those delivered via over-the-air programming.[42][43][44] Developed in partnership with the University of Michigan Transportation Research Institute and the Southwest Research Institute, and in collaboration with stakeholders in industry, academia, and government, Uptane modifies the TUF design to meet the specific security needs of the automotive industry. These needs include accommodating computing units that vary greatly in terms of memory, storage capability, and access to the Internet, while preserving the customizability manufacturers need to design cars for specific client usage.[45] To date, Uptane has been integrated into OTA Plus and ATS Garage, two over-the-air software update products from Advanced Telematic Systems, and is a key security component of the OTAmatic program created by Airbiquity.[46][47] The Airbiquity project was honored with a BIG Award for Business in the 2017 New Product Category in January 2018, and Popular Science magazine named Uptane one of the top 100 inventions for 2017.[48] The first standard volume issued for the project, entitled IEEE-ISTO 6100.1.0.0 Uptane Standard for Design and Implementation, was released on July 31, 2019.[49] Uptane is now a Joint Development Foundation project of the Linux Foundation, operating under the formal title of Joint Development Foundation Projects, LLC, Uptane Series.

In 2016, Cappos introduced in-toto,[50] an open metadata standard that provides documentation of the end-to-end security of a software supply chain. The framework gathers both key information and signatures from all who can access a piece of software through the various stages of coding, testing, building and packaging, thus making transparent all the steps that were performed, by whom and in what order. By creating accountability, in-toto can prevent attackers from either directly introducing malicious changes into the code, or from altering the metadata that keeps the record of those changes along the supply chain.[51] in-toto has collaborated with open source communities such as Docker and OpenSUSE. Datadog utilizes both in-toto and TUF.[52] In December 2020, the framework released its first major version.

While working on in-toto, Cappos and the SSL research group identified metadata manipulation as a new threat against Version Control Systems like Git. His team has developed several new approaches to address this problem, including a defense scheme that mitigates these attacks by maintaining a cryptographically-signed log of relevant developer actions.[53] By documenting the state of the repository at a particular time when an action is taken, developers are given a shared history, so irregularities are easily detected. One recent accomplishment in this research arena is Arch Linux integrating a patch to check for invalid tags in git into the next release of its pacman utility.[54] More recently, Cappos and his collaborators have focused on development of a browser extension that can ensure users of convenient web-based hosting services, such as GitHub or GitLab, that the server will faithfully carry out their requested actions.

Another Cappos project, developed in 2014, introduced a method to make passwords for databases harder to crack. PolyPasswordHasher,[55] is a secure scheme that interrelates stored password data, forcing hackers to crack passwords in sets.[56][57] By making it significantly harder for attackers to figure out the necessary threshold of passwords needed to gain access, PolyPasswordHasher-enabled databases become very difficult to breach. PPH is currently used in several projects, including the Seattle Clearinghouse and BioBank. Implementations are available for seven languages, including Java,[58] Python,[59] C,[60] and Ruby.[61]

• Prof. Justin Cappos, New York University profile page
• Justin Cappos, New York University Tandon School of Engineering profile page
• Secure Systems Laboratory website