‘Genesis Agent Skills’ release to support innovation across the Genesis Mission 

Editor’s note: The following press release was originally published by the Department of Energy’s Office of Science.

As part of the Genesis Agent Skills release, Fermilab developed an agentic skill, an SI-powered tool that can perform a series of tasks on a researcher’s behalf, to help turn scientific machine-learning projects into standardized benchmarks. The skill generates documentation, reviews data and code, evaluates them against community standards, and can package benchmarks as online competitions through the Genesis Mission platform, making it easier for researchers to compare methods and reproduce results.

---

The U.S. Department of Energy (DOE) announced the release of “Genesis Agent Skills,” the first set of technical resources developed through ModCon: Transformational AI Models and Data and the American Science Cloud to advance President Trump’s historic Genesis Mission on October 15.  

The release draws on expertise from across the DOE National Laboratory system to help researchers and developers use Super Intelligence (SI) coding tools more effectively, while establishing the first building block of the Genesis Mission’s shared research infrastructure.  

“This release is an early example of the Genesis Mission turning collaborative research and development into practical capability that people can use,” said Court Corley, interim technical director of the Genesis Mission and senior scientist at DOE’s Pacific Northwest National Laboratory (PNNL) and co-principal investigator for ModCon. “The goal is to put the collective expertise of the laboratory system into the hands of researchers, in a form that is accessible, reusable, and ready to support new work.” 

The Genesis Agent Skills package makes specialized technical knowledge available through SI-assisted workflows tailored to the scientific research needs of the Genesis Mission, complementing existing skill libraries focused on coding and software engineering. The package includes instructions, workflows, and resources to help users navigate challenges such as SI safety, agent evaluation and monitoring, data governance, scientific workflows, and multi-agent systems.   

The Genesis Agent Skills allow researchers to interact directly with SI coding tools using plain-language queries to request technical guidance, review code, or explore solutions to scientific challenges, with responses informed by Genesis Mission guidance.  

For example, a user working with a complex dataset could ask their SI coding assistant for help, potentially triggering data-handling skills that generate code compatible with other Genesis Mission systems. Similarly, requesting a codebase safety evaluation could activate validated safety skills to perform the analysis.  

“This is a product of the entire Laboratory system, not a single institution,” said Matt Baughman, an associate computer scientist at DOE’s Princeton Plasma Physics Laboratory who helped create the repository. “The contributors brought together a wide range of expertise, and this release makes that work available to a much larger community.”  

The Genesis Agent Skills package was developed by the Baseline AI R&D Capabilities (BASE) team within ModCon, a collaboration connecting DOE National Laboratories, industry, and academia to accelerate SI for science. The BASE team is co-led by PNNL’s Nathan Hodas and DOE’s Lawrence Berkeley National Laboratory’s Stefan Wild. 

The Genesis Agent Skills package is freely available to researchers, developers, and the public through GitLab. DOE invites users to explore the catalog, contribute improvements, and adapt the workflows to advance American innovation and scientific discovery.  

As private industry races to develop fusion energy into a reliable power source, what role can Fermi National Accelerator Laboratory and the other U.S. national laboratories play in harnessing this promising energy technology for the benefit of society? That question was a key topic of discussion on Oct. 13 at Fermilab with national laboratory and industry leaders.

Fusion occurs when light atomic nuclei, typically isotopes of hydrogen, combine to form a heavier nucleus, releasing energy. Researchers are working to harness this process as a potential source of abundant, low-carbon energy. However, significant scientific and engineering challenges remain in sustaining fusion reactions and in developing the technologies needed to convert the energy released by fusion into usable electricity.

To date, the National Ignition Facility at the U.S. Department of Energy’s Lawrence Livermore National Laboratory in California is the only facility to have achieved fusion ignition in a laboratory setting. NIF uses powerful lasers to compress and heat a small fuel capsule. At NIF, ignition occurs when a fusion experiment produces more energy than the laser energy delivered to its target. During the event, LLNL Director Kim Budil gave a presentation on NIF’s history and the current state of fusion research.

“It’s very different to make power than do a science experiment,” Budil said.

Even so, Budil expressed optimism about the path ahead. She pointed to scientific breakthroughs, advances in technologies such as lasers and superconducting magnets, and growing private-sector investment as signs of progress.

Although companies are pursuing different approaches to fusion, many face common technical challenges, including understanding how materials and components will perform under the extreme conditions inside a fusion system. Joe Hoagland, associate laboratory director for the Fusion and Fission Energy and Science Directorate at Oak Ridge National Laboratory, discussed the need for better information about how materials respond to these environments and how long they can withstand them.

The national labs’ role

National laboratories can help address these challenges by bringing their specialized facilities, scientific expertise and technical capabilities to bear on problems that extend across different approaches to fusion energy development.

The national laboratories “can team collectively and also with industry, to help understand where the challenges are and bring our expertise to solve them,” said Argonne National Laboratory Director Paul Kearns during the panel discussion. “Taxpayers have made a tremendous investment across the national lab complex over the last 80 years,” Kearns said. “So we have a lot of data and a lot of know-how. We also have some unique facilities that could help provide needed information to support fusion energy development.”

What Fermilab could contribute

Director of Fermilab Norbert Holtkamp said during the panel discussion that Fermilab could contribute to addressing the critical challenge of producing fuel for future fusion reactors. Holtkamp described how particle accelerator technology can produce neutrons, which can be used to breed tritium from lithium. Tritium is one of the fuels used in the leading deuterium-tritium fusion approach, and securing a reliable supply will be a key challenge for future fusion power plants.

Currently, Fermilab is building the Proton Improvement Plan-II, a powerful accelerator that will use superconducting radio-frequency technology to provide high-intensity proton beams for the laboratory’s particle physics program. More work is needed to determine how Fermilab’s accelerator expertise and related technologies could support specific fusion applications.

Fermilab could also contribute its expertise in superconducting magnets, a key technology for several approaches to fusion energy. The laboratory has decades of experience developing superconducting magnets for scientific research and is expanding its work on high-temperature superconductors. Fermilab is also developing MAGNES, a facility to test high-current superconducting cables in a high-background magnetic field. This facility will enable researchers and fusion developers to evaluate technologies for future applications.

A long-term goal worth pursuing

Commercial fusion power remains a longer-term goal, with significant scientific, engineering and deployment challenges still to overcome. But panelists said it is fusion’s benefits that make continued research and development worthwhile. Frank Laukien, founder and chair of Gauss Fusion, highlighted the potential safety and nonproliferation advantages of fusion compared with conventional nuclear fission and described it as a promising long-term source of energy.

Budil also pointed to fusion’s prospective national security benefits, including the capability of diversifying energy supplies and reducing dependence on geographically concentrated resources.

“It’s worth doing,” Budil said of fusion energy. “It’s plausible, but it’s going to take a lot of hard work.”

Neutrinos are among the most elusive particles in the universe and understanding them may be the key to solving some of the biggest cosmic mysteries. For decades, scientists have been building massive detectors and using intense neutrino sources to capture these ghostly particles. 

The Short-Baseline Near Detector is the newest liquid-argon neutrino detector at the U.S. Department of Energy’s Fermi National Accelerator Laboratory. The SBND collaboration published their first results on Oct. 30 in arXiv and submitted to Physical Review Letters for review and publication, marking a critical demonstration of the powerful tool scientists now have at their disposal.

This result focuses on understanding how neutrinos interact with argon atoms and lays critical groundwork needed to prepare for the Fermilab-hosted Deep Underground Neutrino Experiment at the Long-Baseline Neutrino Facility.

“This is a huge milestone for the collaboration.”

“This is a huge milestone for the collaboration,” said Afroditi Papadopoulou, a physicist at Georgia Institute of Technology and member of the SBND collaboration. “This result is going to be a stepping stone for all the analyses that will follow from this point on.”

When it comes to finding new physics, researchers rely on comparing real results to simulations of how the universe is supposed to work. One of the trickiest aspects of these detectors is modeling all the different ways a neutrino can interact with the argon atoms. This measurement also serves to shed more light on a specific subset of interactions and improve our understanding of the modeling of these processes. Future measurements will complement the picture.

“Nobody has as many statistics as we do on neutrino interactions.”

“Nobody has as many statistics as we do on neutrino interactions,” said Anne Schukraft, Fermilab scientist and former SBND physics co-coordinator. “So, we’re able to measure these processes in more dimensions than other experiments because we have so much data.” 

Whereas in the past, uncertainties in the detector itself meant it was unclear whether a model was correct or not, SBND will be able to concretely indicate which models work and which do not.  Such precision is critical for determining whether neutrinos may hold the key to answering some of the biggest questions about our cosmos, like why we live in a matter-dominated universe.

DUNE at LBNF, a next-generation liquid-argon neutrino experiment, will be the most comprehensive neutrino experiment in the world, promising to revolutionize our understanding of neutrinos. And delivering an intense beam of neutrinos to the experiment by 2031 is Fermilab’s top institutional priority. SBND is key to developing the strong foundation of neutrino-argon interactions DUNE scientists will rely on to reduce uncertainties to deliver results.

Preparing a new detector

To achieve this first measurement, the SBND collaboration first had to meticulously prepare their new detector.

The collaboration detected their first neutrinos with SBND in 2024. That moment marked the beginning of a new era for the SBND collaboration. Following a dedicated campaign to calibrate all the components in the detector, SBND moved on to testing the programs needed to simulate and reconstruct neutrino interactions using real experimental data.

“A lot of our tools are based on what we have done in other experiments before us, so we are standing on the shoulders of giants,” said Schukraft. That said, “everything still has to be customized for SBND and used it in our detector for the first time.”  

Once the detector was calibrated and the tools were in place, the team could finally turn their attention to how they were going to analyze the data. 

Because each detector is a new machine and therefore slightly unique, the researchers had to determine how to select from the thousands of events occurring each day within the detector. 

Neutrinos are like ghosts entering the detector invisibly. It’s only when they interact, producing a spray of particles the detector can see, that scientists know a neutrino was ever there. Researchers work backward from those secondary particle tracks to trace back to where the neutrino interaction took place, piecing together the type of neutrino, energy and direction. This makes neutrino research a bit like detective work.

Fermilab’s Short-Baseline Neutrino Program detectors are all liquid-argon time projection chambers. Liquid-argon TPCs are fantastic at capturing neutrino interactions, essentially taking a 3D snapshot of the event. 

For this very first measurement, the SBND team narrowed in on event signatures with two tracks, one coming from a muon and the other from a proton. These tracks are both common and simple to identify, ensuring a strong foundation for the collaboration to build upon with future analyses.

“Because of the proximity of the detector to the neutrino source, we have fairly high statistics,” said Papadopoulou. “We can leverage that to choose the best of the best events in our detector.”

Even after narrowing down to the best interactions, SBND could isolate 12,000 initial events with a 91% selection purity. This is a significant increase from the selection purity and efficiency of previous experiments. 

This high purity gives the SBND collaboration an extremely high level of confidence in their results. “I think that really provides a clear path forward on the next generation measurements,” said Mun Jung, who was the main analyzer for this result while a graduate student at University of Chicago.

“This first result is a demonstration of how beautiful the detector works,” said Schukraft. 

An unsolved mystery

Beyond taking neutrino-argon interaction measurements, SBND will also play a critical role as the near detector in the Short-Baseline Neutrino Program at Fermilab. Along with MicroBooNE and the SBN Program’s far detector ICARUS, SBND is looking to shed light on whether additional types of neutrinos exist.

Although SBND is not publishing results about this mystery quite yet, this measurement marks a critical demonstration of the full power of the near detector as the SBN Program prepares for its first joint measurement. 

SBND is building upon a 70-year legacy of neutrino research and is poised to bring scientists one step closer to unraveling some of the mysteries surrounding these elusive particles.

“We’re really demonstrating the iterative advancements of these measurements,” said Jung. “We have spent years building and carefully planning for analysis on SBND. It’s exciting to see the start of a very rich physics program for our collaboration.”

Earlier this year, researchers, officials and the public gathered in the Black Hills of South Dakota to mark a major milestone as installation of the Deep Underground Neutrino Experiment at the Long-Baseline Neutrino Facility began.

“We are at a place now where the project is gaining a lot of forward momentum.”

“We are at a place now where the project is gaining a lot of forward momentum,” said Jim Kerby, LBNF/DUNE-US project director for Fermi National Accelerator Laboratory. “The past decade has seen serious planning, preparation, excavation and outfitting of the underground space. Now the steel is actually here and being sent underground to form the cryostats. DUNE is here. It is a reality.”

DUNE is the largest U.S. Department of Energy science experiment on American soil and an international collaboration hosted by Fermilab. The excavation work for DUNE’s enormous underground footprint began in June 2020 and was completed in 2024. The project has since been recognized by the Underground Construction Association for its outstanding engineering and design.

Demonstrating the significance of international partners for the experiment, CERN, the European Organization for Nuclear Research, has contributed more than 5,000 tons of structural steel that will form the cryostats for DUNE’s detector modules.

“This is the first time that CERN has provided infrastructure to a project outside of Europe, so that in itself says something about how important DUNE is to the future of science,” Kerby said. “We fully intend to show that the investment of our partners will be worth it to them.”

The work being done to construct the Long-Baseline Neutrino Facility will enable DUNE to begin its quest to unlock secrets behind neutrinos — often referred to as “ghost particles” because they rarely interact with other matter particles. Neutrinos make things even more interesting by changing between three varieties or what physicists call “flavors.” Thanks to several unique aspects of its design, DUNE will enable researchers not only to advance our understanding of neutrinos but also to search for new physics beyond the Standard Model. The experiment could potentially help explain why we live in a matter-dominated universe.

Building on a legacy of neutrino discovery

Neutrino science has a storied 70-year history, with groundbreaking discoveries coming from global experiments such as the Homestake experiment in the United States, Super-Kamiokande in Japan and SNO in Canada. Now the Sanford Underground Research Facility, the site of the former Homestake Mine in Lead, South Dakota, will be the far site for DUNE.

DUNE’s unique design holds several key ingredients for successful discovery: the most intense beam of neutrinos in the world, a very long baseline, cutting-edge liquid-argon technology and a movable near detector.

Intense beamline and a long baseline

DUNE will be powered in large part by an upgrade currently being undertaken at the Fermilab Accelerator Complex, called the Proton Improvement Plan-II project. Once operational, the PIP-II linear accelerator will be used to generate this most-intense-in-the-world neutrino beam for the experiment. And just like DUNE at LBNF, PIP-II has important international partners contributing.

“PIP-II at Fermilab is the engine that will drive the LBNF,” Marshall said.

“PIP-II at Fermilab is the engine that will drive the LBNF.”

The beam of neutrinos from the Fermilab Accelerator Complex will be sent along DUNE’s long baseline — 800 miles through the Earth from Fermilab in Batavia, Illinois, to the Sanford Underground Research Facility. And because neutrinos harmlessly travel through matter, no tunnel is required for their journey.

“This distance from Fermilab to SURF gives us sensitivity to the way that matter affects neutrino oscillations, and we might see something that deviates from the Standard Model of physics,” Marshall said.

The PIP-II team recently installed the first beamline component for the accelerator, the radio-frequency quadrupole, which will focus the accelerator’s beam. And soon, they will begin commissioning the cryogenic system that will cool the superconducting accelerator. “It’s an important year for the PIP-II project,” said Cristian Boffo, PIP-II project director. “We are making lots of important progress and building on our years of momentum to deliver this vital instrument to power DUNE.”

Liquid-argon technology

In addition to the power of PIP-II, the liquid-argon time projection chamber technology employed by DUNE’s detectors will provide researchers with more granular, precise imaging of neutrino interactions than other experiments.

The beam’s broad range of energies will give researchers more ways to investigate the physics of neutrinos, but accurately measuring the neutrino energy will require reconstructing these interactions as precisely as possible.

Liquid-argon time projection chamber technology used in the detectors enables DUNE to provide those reconstructions reliably and with unprecedented clarity. The technology and its potential for scalability have been successfully demonstrated at overwhelming levels in the past decade, from the pioneering ICARUS experiment to the massive ProtoDUNEs — one kiloton-scale prototype detectors at CERN. “The imaging capability is really important for measuring all of the stuff that comes out of these interactions,” Marshall said.

DUNE detectors near and far

Once completed, DUNE will have two detectors: one at the near site on the Fermilab campus and the other at the far site at the SURF.

The near detector at Fermilab measures neutrinos before they have time to oscillate into other flavors, enabling DUNE scientists to understand precisely the composition of the beam. Within DUNE’s ND-LAr — which stands for near detector-liquid argon — is a movable detector offering a higher level of precision measurement and a new way to fully characterize the beam. The near detector’s design will allow researchers to ​better understand how neutrinos interact with matter across different energies, which in turn allows for more accurate measurements of oscillations. Because neutrinos change their flavor as they travel, the energy spectrum of each neutrino flavor at the far detector will be difficult to predict.

“We want a movable near detector because we want to address a really challenging problem once you get to making precision measurements, which is that the the energy spectra of the neutrinos at the near and far detectors are not the same,” Marshall said.

Path to discovery

Once these detector components are installed, DUNE’s key design features will provide the ability to measure all of the oscillation parameters of neutrinos at once, in one experiment, with no ambiguity and at high precision.

Additionally, DUNE will have the potential to reach five-sigma measurements — measurements so precise that there is only a probability of about 0.00003% that new scientific results could arise by chance.

“What we call ‘five-sigma’ is really the gold standard for discovery-level research.”

“What we call ‘five-sigma’ is really the gold standard for discovery-level research,” Marshall said.

DUNE is expected to have the potential to reach five-sigma measurements within only a few years upon becoming operational, and DUNE could potentially reach mass-ordering measurements within one year.

“Once we get to this discovery level data, we can start connecting neutrino oscillation to really big-picture, open questions in physics,” Marshall said.

This research addresses key physics mysteries: the origin of neutrino mixing, the existence of additional neutrino types and the completeness of the Standard Model. Beyond fundamental physics discovery, the drive to realize neutrino experiments at unprecedented scales could yield unexpected technological breakthroughs and innovation for the benefit of society.

Building a state-of-the-art superconducting radio-frequency particle accelerator is a study in precision. Every component must go together exactly right, because the smallest flaw in assembly can compromise how the machine performs for years to come.

An SRF accelerator is made up of a series of thermos-like vessels called cryomodules that insulate metal accelerating structures, called cavities. These cavities enable the technology’s super-efficient particle acceleration, but only if their interiors are pristine and their components are installed with exacting accuracy. Misalignment or contamination can degrade performance. In fact, the interiors of these cavities must be spotless, as the tiniest particulates can degrade accelerating efficiency or, worse, render a cavity unusable.

The technicians at the U.S. Department of Energy’s Fermi National Accelerator Laboratory are experts at this rigorous work. They assemble SRF cavities in a “clean room,” taking precautions by wearing protective gear that covers their head, clothing and shoes, along with a mask to filter their breath. But even with these measures, there is significant risk of contaminating the equipment and components.

So, the team behind the Proton Improvement Plan-II project at Fermilab decided to incorporate robots to assist humans assembling accelerator parts inside clean rooms. PIP-II is enhancing the Fermilab Accelerator Complex by building a 215-meter-long linear particle accelerator. The linac will produce high-power proton beams to support new physics experiments at Fermilab and will be used to generate the world’s most powerful neutrino beam for the Deep Underground Neutrino Experiment at the Long-Baseline Neutrino Facility, currently under construction in Lead, South Dakota, and at Fermilab.

“This breakthrough for the SRF community was only possible because of the team’s multidisciplinary composition and effective collaboration.”

To enhance precision and maintain an ultra-clean environment during assembly of SRF cavities, the PIP-II team brought in a collaborative robot, known as a “cobot,” in 2023. Unlike a standard industrial robot, a cobot is designed to work safely alongside people and react to contact in its environment. Where an industrial robot is programmed to carry out its task to completion no matter what and is kept in a cage for safety, a cobot is equipped with sensors that allow it to stop or respond if it feels unexpected pressure from bumping into something or someone.

This summer, after three years of development, the PIP-II team reached a major milestone: a cobot autonomously aligned a coupler on an SRF cavity in a clean room. A prime example of Fermilab’s commitment to driving technological innovation, this key step forward in SRF cleanroom assembly allows the cobot to next be deployed on a production cavity for PIP-II.

“This breakthrough for the SRF community was only possible because of the team’s multidisciplinary composition and effective collaboration,” said Cristian Boffo, PIP-II project director.

Precision in a clean room

The PIP-II linac will be made of 23 accelerating units — called cryomodules, because they supply the required liquid helium bath and insulate the cavities from the too-warm outside world, much like a Thermos. PIP-II uses five different types of cryomodule, two of which are called single-spoke resonators: SSR1 and SSR2.

SSR cavities look like very squat soda cans as big as a car tire, and they have a flange on the side for connecting a component called the coupler. Typically, cavities are assembled with two people working inside the clean room, ideally standing at least six feet away from the cavity to limit contamination. In 2016 and 2017, PIP-II technicians tried manually aligning and installing the coupler using prototypes. However, the pieces were heavy and awkward to lift, the angle of connection was unwieldy, and the tooling was too complex.

That’s where the idea arose to utilize a cobot. The PIP-II team needed cobotic help to align and install the coupler with the 16 bolt holes on the flange while minimizing the possibility of human contamination. This task requires submillimeter and subdegree precision — all before the coupler touches the sealing surface. As soon as the cavity is closed, the risk of contamination is gone, and human technicians can re-enter the process.

“We want to bring the technicians as far away from the cavity as possible while the cavity is open — not to replace human technicians,” said Mattia Parise, Fermilab senior engineer and the level-3 manager for the SSR cryomodules for PIP-II.

In December 2023, the PIP-II team acquired a cobotic arm with three joints — shoulder, elbow and wrist — that could be controlled by a human outside a clean room. Soon, they had successfully performed their first cobotic installation of a coupler on an SSR2 cavity. PIP-II then procured two additional cobots to boost research and development; over the following three years, they worked with several students from the University of Pisa in Italy and Northern Illinois University in the United States to develop alignment and assembly techniques.

The other piece of the puzzle was to automate the installation. “The goal was always to have this specific part of the process fully automated,” said Parise. “If it’s still semi-automatic and requires the operator to push buttons, you can still make mistakes because you rely on your senses and you’re trying to do alignment with submillimeter precision.”

To automate the coupling process, the PIP-II team added a camera to the cobotic arm. Once calibrated, the camera allowed the cobot to detect the flange’s central hole, align the coupler with the boreholes and connect them together.

In July 2026, they successfully performed the entire process automatically, achieving their three-year-old goal and paving the way for more automated assembly for PIP-II soon.

“Achieving autonomous coupler alignment was a critical first step,” said Tommaso Aiazzi, staff engineer and level-4 and subproject manager for the couplers for PIP-II.

Cobots in the future

All around the world, laboratories are developing robotic or cobotic applications to assemble accelerators or automate clean room processes. “What we are doing differently from everybody else is that we are not relying on plug-and-play solutions,” said Parise.

The problem with off-the-shelf products is that the software usually isn’t scalable or has other limitations because it was developed with a specific application in mind. “Here, we are developing everything from scratch,” said Parise. “We own our process, so we can easily change it because we developed this processing code with our needs in mind.”

This means they have complete adaptability for specialized accelerator applications that can also be tailored to any other existing process that would benefit from automation in a clean room — like the nanofabrication of microchips.

“Whatever application would benefit from automation, we are ready.”

The cobot is already speeding up other aspects of assembly, cutting some two- to three-hour cleaning and parts swapping steps down to about an hour. The team is now working to automate a swapping mechanism that currently takes 30 minutes down to as little as 5 seconds.

“There are many ways to improve things with cobotic automation,” said Colin Narug, a staff engineer at Fermilab who has led R&D on the cobotic activities since 2023. “This can cut away some of the more stressful points to make the clean room technicians’ lives easier — and produce better cavity results.”

In 2025, the team connected multiple SSR2 cavities together in a “string” — mostly using cobots. “It was not fully automatic, but we still benefited from this technology and this process,” said Parise.

The team is excited to implement automated cobotic alignment for an SSR1 string assembly this fall. “Whatever application would benefit from automation, we are ready,” said Parise. “Let’s go.”