pith. sign in

arxiv: 2605.24002 · v1 · pith:BO3GCEO3new · submitted 2026-05-18 · ⚛️ physics.chem-ph · cond-mat.mtrl-sci· cs.AI· physics.comp-ph

Harnessing AtomisticSkills for Agentic Atomistic Research

Bowen Deng , Bohan Li , Matthew Cox , Hoje Chun , Juno Nam , Artur Lyssenko , Sathya Edamadaka , Jurgis Ruza
show 15 more authors
Xiaochen Du Nofit Segal Jesus Diaz Sanchez Mingrou Xie Ty Perez Yu Yao Miguel Steiner Sauradeep Majumdar Charles B. Musgrave III Anirban Chandra Abhirup Patra Detlef Hohl Connor W. Coley Ju Li Rafael G\'omez-Bombarelli
This is my paper

Pith reviewed 2026-06-30 17:55 UTC · model grok-4.3

classification ⚛️ physics.chem-ph cond-mat.mtrl-scics.AIphysics.comp-ph
keywords AtomisticSkillsAI agentscomputational materials sciencemachine learning interatomic potentialsautonomous workflowsmolecular simulationvirtual screeningcatalyst design
0
0 comments X

The pith

AtomisticSkills gives AI coding agents over 100 modular skills to run full atomistic research campaigns.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces AtomisticSkills as a framework that breaks down scientific workflows in computational materials science and chemistry into a hierarchy of more than 100 human-curated skills and tools. These skills cover database access, thermodynamic modeling, MLIP and DFT simulations, and other tasks, allowing general-purpose AI agents to orchestrate them in plug-and-play fashion. Demonstrations include successful execution of generative electrolyte design, MOF screening for CO2 capture, MLIP benchmarking, drug virtual screening, XRD analysis, and catalyst screening. If the approach works as described, agents could manage multi-stage research projects with reduced need for task-specific human coding.

Core claim

AtomisticSkills is an open-source harness framework that empowers general-purpose AI coding agents to conduct atomistic research by hierarchically decomposing workflows into modular, extensible skills and tools, integrating more than 100 multidisciplinary capabilities including database access, thermodynamics and kinetics modeling, and simulation engines using MLIPs and DFT, and validating this through robust orchestration in generative design of Li-ion solid-state electrolytes, high-throughput screening of metal-organic frameworks for CO2 capture, autonomous MLIP benchmarking and fine-tuning, multi-stage structure-based virtual screening for drug design, multimodal X-ray diffraction pattern

What carries the argument

The AtomisticSkills harness, which decomposes workflows hierarchically into agent skills and tools for plug-and-play use.

If this is right

  • Agents can execute generative materials design tasks like creating Li-ion electrolytes using the provided skills.
  • High-throughput screening campaigns for applications such as CO2 capture or drug candidates become executable by agents without per-task custom code.
  • Autonomous benchmarking, fine-tuning of MLIPs, and multimodal data analysis like XRD patterns integrate into agent-driven workflows.
  • Catalyst screening and similar kinetic modeling tasks can be orchestrated across simulation engines by the same agent infrastructure.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The skill library could be expanded to new domains by adding targeted modules, allowing agents to tackle research outside the six demonstrated campaigns.
  • Chaining multiple AtomisticSkills campaigns might enable longer discovery loops where agents iterate on their own outputs.
  • Similar hierarchical skill decomposition could apply to experimental automation if skills for lab equipment control are added.

Load-bearing premise

The curated skills and their integration are complete enough that general AI agents can finish complex multi-stage campaigns without frequent human fixes for missing cases or errors.

What would settle it

An experiment in which a general-purpose AI agent equipped only with AtomisticSkills attempts a new unseen multi-stage campaign, such as screening for a different class of catalysts, and either completes it end-to-end or requires substantial additional human-written code or interventions.

Figures

Figures reproduced from arXiv: 2605.24002 by Abhirup Patra, Anirban Chandra, Artur Lyssenko, Bohan Li, Bowen Deng, Charles B. Musgrave III, Connor W. Coley, Detlef Hohl, Hoje Chun, Jesus Diaz Sanchez, Ju Li, Juno Nam, Jurgis Ruza, Matthew Cox, Miguel Steiner, Mingrou Xie, Nofit Segal, Rafael G\'omez-Bombarelli, Sathya Edamadaka, Sauradeep Majumdar, Ty Perez, Xiaochen Du, Yu Yao.

Figure 1
Figure 1. Figure 1: FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2 [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (a) illustrates the number of skills we identified from each article in npj Computational Materials. The distribution indicates that typical computational stud￾ies rely on fewer than 10 interconnected skills. Crucially, evaluating these operations against our repository reveals a notably high degree of functional coverage [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (b) shows the distribution of energies above the hull (Ehull) calculated by the UMA-s-1p2-omat [54], where candidates with lower Ehull are thermodynamically more stable. The inset schematic plot illustrates how the Ehull and stable chemical potential (µ) ranges are cal￾culated from the 0 K computational convex hull phase diagram [53]. The top candidates then underwent MCP￾tool-triggered multi-temperature M… view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6 [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7 [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
read the original abstract

Computational materials science and chemistry span vast knowledge domains and fractured software ecosystems. Although large language models (LLMs) have demonstrated research capabilities, scaling monolithic agents to manage the rigor and complexity of atomistic research remains a challenge. Here, we introduce AtomisticSkills, an open-source harness framework that empowers general-purpose AI coding agents to conduct atomistic research across materials science, chemistry, and drug discovery. By hierarchically decomposing scientific workflows into agent skills and tools, AtomisticSkills provides agents with modular, extensible, and plug-and-play research capabilities. The framework integrates more than 100 human-curated multidisciplinary skills, including database access, thermodynamics and kinetics modeling, and diverse simulation engines employing machine learning interatomic potentials (MLIPs) and density functional theory (DFT). We validate its functional coverage against scientific literature and demonstrate robust orchestration capabilities across diverse scientific campaigns: generative design of Li-ion solid-state electrolytes, high-throughput screening of metal-organic frameworks for CO2 capture, autonomous MLIP benchmarking and fine-tuning, multi-stage structure-based virtual screening for drug design, multimodal X-ray diffraction pattern analysis, and screening of Fe-oxide catalysts for oxygen evolution reaction. AtomisticSkills provides a critical agent infrastructure towards building fully autonomous AI scientists.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript introduces AtomisticSkills, an open-source framework that hierarchically decomposes atomistic research workflows into more than 100 human-curated modular skills and tools (including database access, thermodynamics modeling, MLIP/DFT simulations) for use by general-purpose AI coding agents. It claims functional coverage validated against literature and demonstrates 'robust orchestration' on six campaigns: generative design of Li-ion solid-state electrolytes, high-throughput MOF screening for CO2 capture, autonomous MLIP benchmarking/fine-tuning, multi-stage virtual screening for drug design, multimodal XRD pattern analysis, and Fe-oxide catalyst screening for the oxygen evolution reaction. The central contribution is positioned as critical infrastructure toward fully autonomous AI scientists.

Significance. If the framework's skills enable reliable, low-intervention autonomous execution of multi-stage campaigns, the work would supply a reusable, extensible harness that lowers the barrier for agentic AI in materials and chemistry research. The open-source release, integration of diverse simulation engines, and coverage of multidisciplinary tasks (from generative design to experimental data analysis) represent practical strengths that could accelerate reproducible agent-based workflows if quantitative performance data were supplied.

major comments (1)
  1. [Abstract] Abstract: The central claim of 'robust orchestration capabilities' across the six listed campaigns (and the broader assertion of providing 'critical agent infrastructure towards building fully autonomous AI scientists') is load-bearing yet unsupported by any quantitative metrics. No success rates, completion fractions, retry counts, human-intervention statistics, failure-mode analysis, or baseline comparisons (e.g., against un-augmented agents) are reported for any campaign, preventing evaluation of the weakest assumption that the >100 curated skills suffice for reliable autonomous operation.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive feedback. We address the single major comment below and commit to revisions that accurately reflect the manuscript's scope.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim of 'robust orchestration capabilities' across the six listed campaigns (and the broader assertion of providing 'critical agent infrastructure towards building fully autonomous AI scientists') is load-bearing yet unsupported by any quantitative metrics. No success rates, completion fractions, retry counts, human-intervention statistics, failure-mode analysis, or baseline comparisons (e.g., against un-augmented agents) are reported for any campaign, preventing evaluation of the weakest assumption that the >100 curated skills suffice for reliable autonomous operation.

    Authors: We agree that the abstract's phrasing of 'robust orchestration capabilities' and positioning as 'critical agent infrastructure towards building fully autonomous AI scientists' is not supported by quantitative metrics such as success rates, intervention counts, or baseline comparisons. The six campaigns serve as illustrative demonstrations of functional coverage, workflow decomposition, and integration with simulation engines, validated qualitatively against literature, rather than as statistical evaluations of autonomous performance. In the revised manuscript we will (1) revise the abstract to state that AtomisticSkills 'enables application to' the listed campaigns and (2) add an explicit Limitations section that acknowledges the lack of quantitative agent-performance benchmarks and identifies this as an important direction for future work. These changes will be incorporated in the next version. revision: yes

Circularity Check

0 steps flagged

No circularity: framework description with no derivations or self-referential fits

full rationale

The paper introduces and describes a software framework (AtomisticSkills) with >100 curated skills for agentic workflows. It lists campaigns and claims functional coverage validated against literature, but contains no equations, parameter fits, predictions, or derivation chains. Claims rest on the existence of the implemented skills and their demonstrated use, which are externally verifiable via the open-source release rather than reducing to self-definition or self-citation. This matches the default expectation of no significant circularity for non-mathematical software/framework contributions.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

The central claim rests on the existence, completeness, and usability of the 100+ skills plus the assumption that general AI agents can orchestrate them effectively; no free parameters, mathematical axioms, or external benchmarks are mentioned in the abstract.

invented entities (1)
  • AtomisticSkills no independent evidence
    purpose: To give general-purpose AI coding agents modular capabilities for atomistic research workflows
    The framework itself is the primary new entity introduced; no independent evidence of its performance is supplied in the abstract.

pith-pipeline@v0.9.1-grok · 5850 in / 1199 out tokens · 31799 ms · 2026-06-30T17:55:38.782258+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

125 extracted references · 18 canonical work pages · 4 internal anchors

  1. [1]

    K. Jun, Y. Sun, Y. Xiao, Y. Zeng, R. Kim, H. Kim, L. J. Miara, D. Im, Y. Wang, and G. Ceder, Lithium superionic conductors with corner-sharing frameworks, Nature Materials21, 924–931 (2022)

    2022

  2. [2]

    C. G. V. d. Walle and J. Neugebauer, First-principles calculations for defects and impurities: Applications to iii-nitrides, Journal of Applied Physics95, 3851–3879 (2004)

    2004

  3. [3]

    B. Qiao, A. Wang, X. Yang, L. F. Allard, Z. Jiang, Y. Cui, J. Liu, J. Li, and T. Zhang, Single-atom cataly- sis of co oxidation using pt1/feox, Nature Chemistry3, 634–641 (2011)

    2011

  4. [4]

    T. A. Halgren, Merck molecular force field. I. Ba- sis, form, scope, parameterization, and performance of MMFF94, Journal of Computational Chemistry17, 490 (1996)

    1996

  5. [5]

    A. M. Ganose, H. Sahasrabuddhe, M. Asta, K. Beck, T. Biswas, A. Bonkowski, J. Bustamante, X. Chen, Y. Chiang, D. C. Chrzan, J. Clary, O. A. Cohen, C. Er- tural, M. C. Gallant, J. George, S. Gerits, R. E. A. Goodall, R. D. Guha, G. Hautier, M. Horton, T. J. Inizan, A. D. Kaplan, R. S. Kingsbury, M. C. Kuner, B. Li, X. Linn, M. J. McDermott, R. S. Mohanakri...

    1944

  6. [6]

    Curtarolo, W

    S. Curtarolo, W. Setyawan, G. L. Hart, M. Jahnatek, R. V. Chepulskii, R. H. Taylor, S. Wang, J. Xue, K. Yang, O. Levy, M. J. Mehl, H. T. Stokes, D. O. Dem- chenko, and D. Morgan, Aflow: An automatic frame- work for high-throughput materials discovery, Compu- tational Materials Science58, 218–226 (2012)

    2012

  7. [7]

    Uhrin, S

    M. Uhrin, S. P. Huber, J. Yu, N. Marzari, and G. Pizzi, Workflows in AiiDA: Engineering a high-throughput, event-based engine for robust and modular computa- tional workflows, Computational Materials Science187, 110086 (2021)

    2021

  8. [8]

    A. Jain, S. P. Ong, W. Chen, B. Medasani, X. Qu, M. Kocher, M. Brafman, G. Petretto, G.-M. Rignanese, G. Hautier, D. Gunter, and K. A. Persson, FireWorks: a dynamic workflow system designed for high-throughput applications, Concurrency and Computation: Practice and Experience27, 5037 (2015)

    2015

  9. [9]

    G. Wang, L. Peng, K. Li, L. Zhu, J. Zhou, N. Miao, and Z. Sun, ALKEMIE: An intelligent computational platform for accelerating materials discovery and design, Computational Materials Science186, 110064 (2021)

    2021

  10. [10]

    Weymuth, J

    T. Weymuth, J. P. Unsleber, P. L. T¨ urtscher, M. Steiner, J.-G. Sobez, C. H. M¨ uller, M. M¨ orchen, V. Klasovita, S. A. Grimmel, M. Eckhoff, K.-S. Csizi, F. Bosia, M. Bensberg, and M. Reiher, SCINE— Software for chemical interaction networks, J. Chem. Phys.160(2024)

    2024

  11. [11]

    M. K. Horton, P. Huck, R. X. Yang, J. M. Munro, S. Dwaraknath, A. M. Ganose, R. S. Kingsbury, M. Wen, J. X. Shen, T. S. Mathis, A. D. Kaplan, K. Berket, J. Riebesell, J. George, A. S. Rosen, E. W. C. Spotte-Smith, M. J. McDermott, O. A. Co- hen, A. Dunn, M. C. Kuner, G.-M. Rignanese, G. Pe- tretto, D. Waroquiers, S. M. Griffin, J. B. Neaton, D. C. Chrzan,...

    2025

  12. [12]

    D. A. Boiko, R. MacKnight, B. Kline, and G. Gomes, Autonomous chemical research with large language models, Nature624, 570–578 (2023)

    2023

  13. [13]

    T. Song, M. Luo, X. Zhang, L. Chen, Y. Huang, J. Cao, Q. Zhu, D. Liu, B. Zhang, G. Zou, G. Zhang, F. Zhang, W. Shang, Y. Fu, J. Jiang, and Y. Luo, A multiagent- driven robotic ai chemist enabling autonomous chemical research on demand, Journal of the American Chemical Society147, 12534–12545 (2025)

    2025

  14. [14]

    Y. Fei, B. Rendy, X. Yang, J. Woo, X. Huang, C. Li, S. Wang, D. Milsted, Y. Zeng, and G. Ceder, Agentic llm reasoning in a self-driving laboratory for air-sensitive lithium halide spinel conductors, arXiv (2026)

    2026

  15. [15]

    Zhang, Z

    Z. Zhang, Z. Ren, C.-W. Hsu, W. Chen, Z.-W. Hong, C.- F. Lee, A. Penn, H. Xu, D. J. Zheng, S. Miao, Y. Huang, Y. Gao, W. Chen, H. Smith, Y. Niu, Y. Tian, Y.-R. Lu, Y.-C. Shao, S. Li, H.-T. Wang, I. I. Abate, P. Agrawal, Y. Shao-Horn, and J. Li, A multimodal robotic platform for multi-element electrocatalyst discovery, Nature , 1–3 (2025)

    2025

  16. [16]

    Chiang, E

    Y. Chiang, E. Hsieh, C.-H. Chou, and J. Riebesell, Llamp: Large language model made powerful for high- fidelity materials knowledge retrieval, Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , 25200–25232 (2025)

    2025

  17. [17]

    Y. Zou, A. H. Cheng, A. Aldossary, J. Bai, S. X. Leong, J. A. Campos-Gonzalez-Angulo, C. Choi, C. T. Ser, G. Tom, A. Wang, Z. Zhang, I. Yakavets, H. Hao, C. Crebolder, V. Bernales, and A. Aspuru-Guzik, El agente: An autonomous agent for quantum chemistry, Matter8, 102263 (2025)

    2025

  18. [18]

    Gustin, L

    I. Gustin, L. M. Calder´ on, J. B. P´ erez-S´ anchez, J. F. Gonthier, Y. Nakamura, K. Panicker, M. Ramprasad, Z. Zhang, Y. Zou, V. Bernales, and A. Aspuru-Guzik, El agente cu´ antico: Automating quantum simulations, arXiv (2025)

    2025

  19. [19]

    Marwitz, A

    T. Marwitz, A. Colsmann, B. Breitung, C. Brabec, C. Kirchlechner, E. Blasco, G. C. Marques, H. Hahn, M. Hirtz, P. A. Levkin, Y. M. Eggeler, T. Schl¨ oder, and P. Friederich, Predicting new research directions in ma- terials science using large language models and concept graphs, Nature Machine Intelligence , 1–10 (2026)

    2026

  20. [20]

    A. M. Bran, S. Cox, O. Schilter, C. Baldassari, A. D. White, and P. Schwaller, Augmenting large language models with chemistry tools, Nature Machine Intelli- gence6, 525–535 (2024)

    2024

  21. [21]

    Ghafarollahi and M

    A. Ghafarollahi and M. J. Buehler, Sciagents: Au- tomating scientific discovery through bioinspired multi- agent intelligent graph reasoning, Advanced Materials 37, 2413523 (2025)

    2025

  22. [22]

    J. Hu, H. Nawaz, Y.-F. Hou, Y. Rui, L. Chi, Y. Chen, A. Ullah, and P. O. Dral, Aitomia: Your intelligent as- sistant for ai-driven atomistic and quantum chemical simulations, arXiv preprint arXiv:2505.08195 (2025)

    work page arXiv 2025

  23. [23]

    Mitchener, A

    L. Mitchener, A. Yiu, B. Chang, M. Bourdenx, T. Nadolski, A. Sulovari, E. C. Landsness, D. L. Barabasi, S. Narayanan, N. Evans, S. Reddy, M. Foiani, A. Kamal, L. P. Shriver, F. Cao, A. T. Wassie, J. M. Laurent, E. Melville-Green, M. Caldas, A. Bou, K. F. Roberts, S. Zagorac, T. C. Orr, M. E. Orr, K. J. Zwezdaryk, A. E. Ghareeb, L. McCoy, B. Gomes, E. A. A...

    2025

  24. [24]

    Zheng, N

    Z. Zheng, N. Rampal, T. J. Inizan, C. Borgs, J. T. Chayes, and O. M. Yaghi, Large language models for reticular chemistry, Nature Reviews Materials10, 369–381 (2025)

    2025

  25. [25]

    Y. Wang, Z. Huang, Z. Ding, R. Liao, Y. Huang, X. Liu, J. Xie, S. Chen, and L. Zhang, Deploy-master: Au- tomating the deployment of 50,000+ agent-ready sci- entific tools in one day, arXiv (2026)

    2026

  26. [26]

    Zhang, S

    L. Zhang, S. Chen, Y. Cai, J. Chai, J. Chang, K. Chen, Z. X. Chen, Z. Ding, Y. Du, Y. Gao, Y. Gao, J. Gao, Z. Gao, Q. Gu, Y. Hong, Y. Huang, X. Fang, X. Ji, G. Ke, Z. Lei, X. Li, Y. Li, R. Liao, H. Lin, X. Lin, Y. Liu, X. Liu, Z. Liu, J. Lu, T. Miao, H. Que, W. Sun, Y. Wang, B. Wu, T. Xue, R. Ye, J. Zeng, D. Zhang, J. Zhang, L. Zhang, T. Zhang, W. Zhang, ...

    work page doi:10.48550/arxiv.2512.20469 2025

  27. [27]

    Ruˇ za and R

    J. Ruˇ za and R. Gomez-Bombarelli, Reasoning-to- simulation: An agentic framework for discovery of elec- trolyte materials, inAI for Accelerated Materials Design - ICLR 2026(2026)

    2026

  28. [28]

    Soleymanibrojeni, R

    M. Soleymanibrojeni, R. Aydin, D. Guedes-Sobrinho, A. C. Dias, M. J. Piotrowski, W. Wenzel, and C. R. C. Rˆ ego, Genius: an agentic ai framework for autonomous design and execution of simulation protocols, Commu- nications Materials7, 115 (2026)

    2026

  29. [29]

    S. G. H. Kumar, Y. Zou, A. Wang, J. Vald´ es- Hern´ andez, T. W. Ko, N. Yue, O. Leng, H. Xu, C. Cre- bolder, A. Aspuru-Guzik, and V. Bernales, El agente s´ olido: A new age(nt) for solid state simulations, arXiv 10.48550/arxiv.2602.17886 (2026)

    work page doi:10.48550/arxiv.2602.17886 2026

  30. [30]

    Chase, LangChain (2022)

    H. Chase, LangChain (2022)

    2022

  31. [31]

    Anthropic, Claude code,https://www.anthropic.com/ claude-code(2025)

    2025

  32. [32]

    Gauthier, Aider: AI pair programming in your termi- nal,https://aider.chat(2023), open-source software, https://github.com/Aider-AI/aider

    P. Gauthier, Aider: AI pair programming in your termi- nal,https://aider.chat(2023), open-source software, https://github.com/Aider-AI/aider

    2023

  33. [33]

    OpenAI, Introducing Codex,https://openai.com/i ndex/introducing-codex/(2025), codex CLI released April 2025; cloud agent research preview May 2025; gen- erally available October 6, 2025

    2025

  34. [34]

    Google, Gemini CLI: Your open-source AI agent,https: //blog.google/technology/developers/introducin g-gemini-cli-open-source-ai-agent/(2025), open- source software,https://github.com/google-gemini/ gemini-cli

    2025

  35. [35]

    Anysphere, Cursor: The AI Code Editor,https://cu rsor.com(2023)

    2023

  36. [36]

    Google, Build with Google Antigravity, our new agentic development platform,https://developers.googleb log.com/build-with-google-antigravity-our-new -agentic-development-platform/(2025)

    2025

  37. [37]

    Codeium, Windsurf Editor,https://windsurf.com (2024), codeium rebranded to Windsurf in April 2025; acquired by OpenAI in May 2025

    2024

  38. [38]

    C. Zhou, H. Chai, W. Chen, Z. Guo, R. Shan, Y. Song, T. Xu, Y. Yang, A. Yu, W. Zhang, C. Zheng, J. Zhu, Z. Zheng, Z. Zhang, X. Lou, C. Zhang, Z. Fu, J. Wang, W. Liu, J. Lin, and W. Zhang, Externalization in llm agents: A unified review of memory, skills, protocols and harness engineering, arXiv 10.48550/arxiv.2604.08224 (2026)

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2604.08224 2026

  39. [39]

    Kapoor, B

    S. Kapoor, B. Stroebl, Z. S. Siegel, N. Nadgir, and A. Narayanan, AI agents that matter, Transactions on Machine Learning Research (2025)

    2025

  40. [40]

    Kapoor, B

    S. Kapoor, B. Stroebl, P. Kirgis, N. Nadgir, Z. S. Siegel, B. Wei, T. Xue, Z. Chen, F. Chen, S. Utpala, F. Nd- zomga, D. Oruganty, S. Luskin, K. Liu, B. Yu, A. Arora, D. Hahm, H. Trivedi, H. Sun, J. Lee, T. Jin, Y. Mai, Y. Zhou, Y. Zhu, R. Bommasani, D. Kang, D. Song, P. Henderson, Y. Su, P. Liang, and A. Narayanan, Holis- tic agent leaderboard: The missi...

    2026

  41. [41]

    Anthropic, Equipping agents for the real world with Agent Skills,https://www.anthropic.com/engine ering/equipping-agents-for-the-real-world-wit h-agent-skills(2025)

    2025

  42. [42]

    Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks

    P. Lewis, E. Perez, A. Piktus, F. Petroni, V. Karpukhin, N. Goyal, H. K¨ uttler, M. Lewis, W.-t. Yih, T. Rockt¨ aschel, S. Riedel, and D. Kiela, Retrieval- augmented generation for knowledge-intensive nlp tasks, arXiv 10.48550/arxiv.2005.11401 (2020)

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2005.11401 2005

  43. [43]

    N. F. Liu, K. Lin, J. Hewitt, A. Paranjape, M. Bevilac- qua, F. Petroni, and P. Liang, Lost in the middle: How language models use long contexts, Transactions of the Association for Computational Linguistics12, 157 (2024)

    2024

  44. [44]

    RULER: What's the Real Context Size of Your Long-Context Language Models?

    C.-P. Hsieh, S. Sun, S. Kriman, S. Acharya, D. Rekesh, F. Jia, Y. Zhang, and B. Ginsburg, Ruler: What’s the real context size of your long-context language models?, arXiv preprint arXiv:2404.06654 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  45. [45]

    Modarressi, H

    A. Modarressi, H. Deilamsalehy, F. Dernoncourt, T. Bui, R. A. Rossi, S. Yoon, and H. Sch¨ utze, No- lima: Long-context evaluation beyond literal matching, inForty-second International Conference on Machine Learning(2025)

    2025

  46. [46]

    K. Hong, A. Troynikov, and J. Huber,Context Rot: How Increasing Input Tokens Impacts LLM Performance, Tech. Rep. (Chroma, 2025)

    2025

  47. [47]

    F. Shi, X. Chen, K. Misra, N. Scales, D. Dohan, E. Chi, N. Sch¨ arli, and D. Zhou, Large language models can be easily distracted by irrelevant context, inProceedings of the 40th International Conference on Machine Learning, ICML’23 (JMLR.org, 2023)

    2023

  48. [48]

    Anthropic, Introducing the Model Context Protocol,ht tps://www.anthropic.com/news/model-context-pro tocol(2024), accessed: 2026-04-28

    2024

  49. [49]

    S. Gao, R. Zhu, P. Sui, Z. Kong, S. Aldogom, Y. Huang, A. Noori, R. Shamji, K. Parvataneni, T. Tsiligkaridis, and M. Zitnik, Democratizing ai scientists using toolu- niverse, arXiv 10.48550/arxiv.2509.23426 (2025)

    work page doi:10.48550/arxiv.2509.23426 2025

  50. [50]

    K-Dense Inc., Scientific agent skills: A comprehensive collection of scientific tools for ai agents (2026), 133 skills covering databases, packages, integrations, and analysis tools

    2026

  51. [51]

    Zhang, L

    L. Zhang, L. Wang, X. Sun, W. Tang, H. Su, Y. Qian, Q. Yang, Q. Li, Z. Tang, H. Sun, Y. Han, Y. Jiang, W. Lou, B. Zhou, X. Wang, L. Bai, and Z. Xie, Molclaw: An autonomous agent with hierarchical skills for drug molecule evaluation, screening, and optimization, arXiv (2026)

    2026

  52. [52]

    Ishimaru, S

    M. Ishimaru, S. Munetoh, and T. Motooka, Genera- tion of amorphous silicon structures by rapid quench- ing: A molecular-dynamics study, Physical Review B 56, 15133–15138 (1997)

    1997

  53. [53]

    W. Sun, S. T. Dacek, S. P. Ong, G. Hautier, A. Jain, W. D. Richards, A. C. Gamst, K. A. Persson, and G. Ceder, The thermodynamic scale of inorganic crys- talline metastability, Science Advances2, e1600225 (2016)

    2016

  54. [54]

    B. M. Wood, M. Dzamba, X. Fu, M. Gao, M. Shuaibi, L. Barroso-Luque, K. Abdelmaqsoud, V. Gharakhanyan, J. R. Kitchin, D. S. Levine, K. Michel, A. Sriram, T. Cohen, A. Das, A. Rizvi, S. J. Sahoo, Z. W. Ulissi, and C. L. Zitnick, Uma: A family of universal models for atoms, arXiv 10.48550/arxiv.2506.23971 (2025)

    work page doi:10.48550/arxiv.2506.23971 2025

  55. [55]

    T. W. Ko, B. Deng, M. Nassar, L. Barroso-Luque, R. Liu, J. Qi, A. C. Thakur, A. R. Mishra, E. Liu, G. Ceder, S. Miret, and S. P. Ong, Materials graph li- brary (matgl), an open-source graph deep learning li- 16 brary for materials science and chemistry, npj Compu- tational Materials11, 253 (2025)

    2025

  56. [56]

    Batatia, P

    I. Batatia, P. Benner, Y. Chiang, A. M. Elena, D. P. Kov´ acs, J. Riebesell, X. R. Advincula, M. Asta, M. Avaylon, W. J. Baldwin, F. Berger, N. Bernstein, A. Bhowmik, F. Bigi, S. M. Blau, V. C˘ arare, M. Ce- riotti, S. Chong, J. P. Darby, S. De, F. D. Pia, V. L. Deringer, R. Elijoˇ sius, Z. El-Machachi, E. Fako, F. Fal- cioni, A. C. Ferrari, J. L. A. Gard...

    2025

  57. [57]

    Barroso-Luque, J

    L. Barroso-Luque, J. H. Yang, F. Xie, T. Chen, R. L. Kam, Z. Jadidi, P. Zhong, and G. Ceder, smol: A python package for cluster expansions and beyond, Journal of Open Source Software7, 4504 (2022)

    2022

  58. [58]

    C. Zeni, R. Pinsler, D. Z¨ ugner, A. Fowler, M. Hor- ton, X. Fu, Z. Wang, A. Shysheya, J. Crabb´ e, S. Ueda, R. Sordillo, L. Sun, J. Smith, B. Nguyen, H. Schulz, S. Lewis, C.-W. Huang, Z. Lu, Y. Zhou, H. Yang, H. Hao, J. Li, C. Yang, W. Li, R. Tomioka, and T. Xie, A generative model for inorganic materials design, Na- ture , 1–3 (2025)

    2025

  59. [59]

    R. Jiao, W. Huang, Y. Liu, D. Zhao, and Y. Liu, Space group constrained crystal generation, inThe Twelfth International Conference on Learning Representations (2024)

    2024

  60. [60]

    G. S. Na, Cross-modality material embedding loss for transferring knowledge between heterogeneous mate- rial descriptors, npj Computational Materials11, 235 (2025)

    2025

  61. [61]

    T. Ueno, H. Hino, A. Hashimoto, Y. Takeichi, M. Sawada, and K. Ono, Adaptive design of an x-ray magnetic circular dichroism spectroscopy experiment with gaussian process modelling, npj Computational Materials4, 4 (2018)

    2018

  62. [62]

    C. Liu, Y. Bai, J. Zhou, Q. Zhao, Y. Yang, H. Chen, and L. Qiao, High-performance bifunctional polarization switch chiral metamaterials by inverse design method, npj Computational Materials5, 93 (2019)

    2019

  63. [63]

    T. Zhao, Y. Zhang, Q. Wei, X. Shi, P. Zhao, L.-Q. Chen, and S. Zhang, Active cell-matrix coupling regulates cel- lular force landscapes of cohesive epithelial monolayers, npj Computational Materials4, 10 (2018)

    2018

  64. [64]

    Y. Zhu, X. He, and Y. Mo, Origin of outstanding sta- bility in the lithium solid electrolyte materials: Insights from thermodynamic analyses based on first-principles calculations, ACS Applied Materials & Interfaces7, 23685–23693 (2015)

    2015

  65. [65]

    C. Chen, D. T. Nguyen, S. J. Lee, N. A. Baker, A. S. Karakoti, L. Lauw, C. Owen, K. T. Mueller, B. A. Bilodeau, V. Murugesan, and M. Troyer, Accelerating computational materials discovery with machine learn- ing and cloud high-performance computing: from large- scale screening to experimental validation, Journal of the American Chemical Society 10.1021/ja...

    work page doi:10.1021/jacs.4c03849 2024

  66. [66]

    Majumdar, S

    S. Majumdar, S. M. Moosavi, K. M. Jablonka, D. On- gari, and B. Smit, Diversifying databases of metal or- ganic frameworks for high-throughput computational screening, ACS applied materials & interfaces13, 61004 (2021)

    2021

  67. [67]

    H. Park, S. Majumdar, X. Zhang, J. Kim, and B. Smit, Inverse design of metal–organic frameworks for direct air capture of co 2 via deep reinforcement learning, Dig- ital Discovery3, 728 (2024)

    2024

  68. [68]

    Kang and J

    Y. Kang and J. Kim, Chatmof: an artificial intel- ligence system for predicting and generating metal- organic frameworks using large language models, Nature Communications15, 4705 (2024)

    2024

  69. [69]

    A. S. Rosen, S. M. Iyer, D. Ray, Z. Yao, A. Aspuru- Guzik, L. Gagliardi, J. M. Notestein, and R. Q. Snurr, Machine learning the quantum-chemical properties of metal–organic frameworks for accelerated materials dis- covery, Matter4, 1578 (2021)

    2021

  70. [70]

    Burner, J

    J. Burner, J. Luo, A. White, A. Mirmiran, O. Kwon, P. G. Boyd, S. Maley, M. Gibaldi, S. Simrod, V. Og- den,et al., Arc–mof: a diverse database of metal-organic frameworks with dft-derived partial atomic charges and descriptors for machine learning, Chemistry of Materi- als35, 900 (2023)

    2023

  71. [71]

    Widom, Some topics in the theory of fluids, The Jour- nal of Chemical Physics39, 2808–2812 (1963)

    B. Widom, Some topics in the theory of fluids, The Jour- nal of Chemical Physics39, 2808–2812 (1963)

    1963

  72. [72]

    Kumar, D

    A. Kumar, D. G. Madden, M. Lusi, K.-J. Chen, E. A. Daniels, T. Curtin, J. J. Perry, and M. J. Zaworotko, Di- rect air capture of co2 by physisorbent materials, Ange- wandte Chemie International Edition54, 14372–14377 (2015)

    2015

  73. [73]

    Samanta, A

    A. Samanta, A. Zhao, G. K. H. Shimizu, P. Sarkar, and R. Gupta, Post-combustion co2 capture using solid sor- bents: A review, Industrial & Engineering Chemistry Research51, 1438–1463 (2012)

    2012

  74. [74]

    Jiang, W

    L. Jiang, W. Liu, R. Wang, A. Gonzalez-Diaz, M. Rojas- Michaga, S. Michailos, M. Pourkashanian, X. Zhang, and C. Font-Palma, Sorption direct air capture with co2 utilization, Progress in Energy and Combustion Science 95, 101069 (2023)

    2023

  75. [75]

    Chen and S

    C. Chen and S. P. Ong, A universal graph deep learn- ing interatomic potential for the periodic table, Nature Computational Science2, 718–728 (2022)

    2022

  76. [76]

    Jacobs, D

    R. Jacobs, D. Morgan, S. Attarian, J. Meng, C. Shen, Z. Wu, C. Y. Xie, J. H. Yang, N. Artrith, B. Blaiszik, G. Ceder, K. Choudhary, G. Csanyi, E. D. Cubuk, B. Deng, R. Drautz, X. Fu, J. Godwin, V. Honavar, O. Isayev, A. Johansson, S. Martiniani, S. P. Ong, I. Poltavsky, K. Schmidt, S. Takamoto, A. P. Thomp- son, J. Westermayr, B. M. Wood, and B. Kozinsky,...

    2025

  77. [77]

    E. C.-Y. Yuan, Y. Liu, J. Chen, P. Zhong, S. Raja, T. Kreiman, S. Vargas, W. Xu, M. Head-Gordon, C. Yang, S. M. Blau, B. Cheng, A. Krishnapriyan, 17 and T. Head-Gordon, Foundation models for atomistic simulation of chemistry and materials, Nature Reviews Chemistry10, 212–230 (2026)

    2026

  78. [78]

    B. Deng, P. Zhong, K. Jun, J. Riebesell, K. Han, C. J. Bartel, and G. Ceder, Chgnet as a pretrained universal neural network potential for charge-informed atomistic modelling, Nature Machine Intelligence , 1–11 (2023)

    2023

  79. [79]

    A. D. Kaplan, R. Liu, J. Qi, T. W. Ko, B. Deng, J. Riebesell, G. Ceder, K. A. Persson, and S. P. Ong, A foundational potential energy surface dataset for mate- rials, arXiv 10.48550/arxiv.2503.04070 (2025)

    work page doi:10.48550/arxiv.2503.04070 2025

  80. [80]

    Perez and R

    T. Perez and R. Gomez-Bombarelli, Self-conditioned denoising for atomistic representation learning, arXiv 10.48550/arXiv.2603.17196 (2026), arXiv:2603.17196

    work page doi:10.48550/arxiv.2603.17196 2026

Showing first 80 references.