pith. sign in

arxiv: 2605.15625 · v1 · pith:3MWLMQAVnew · submitted 2026-05-15 · 💻 cs.AI · cond-mat.soft

ColPackAgent: Agent-Skill-Guided Hard-Particle Monte Carlo Workflows for Colloidal Packing

Lijie Ding , Changwoo Do This is my paper

Pith reviewed 2026-05-20 19:24 UTC · model grok-4.3

classification 💻 cs.AI cond-mat.soft
keywords agent frameworkMonte Carlo simulationcolloidal packingMCP toolsagent skillhard-particle Monte Carloworkflow automationLLM agent
0
0 comments X

The pith

Pairing a domain Python package with MCP tools and a portable agent skill turns a simulation toolkit into an agent-assisted research workflow.

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

The paper establishes ColPackAgent as a system that lets agents autonomously execute Monte Carlo simulations for colloidal packing instead of merely describing the steps. It achieves this through an MCP tool server that exposes a custom Python package wrapping hard-particle Monte Carlo methods and an agent skill that enforces a four-stage workflow contract. General-purpose LLM agents without these dedicated components typically stop at textual descriptions rather than carrying out the full simulation sequence. A reader would care because colloidal packing studies underpin research on phase behavior, self-assembly, and materials design, and the approach makes such simulations more accessible through structured agent guidance.

Core claim

ColPackAgent autonomously runs Monte Carlo simulations of colloidal packing through a Model Context Protocol (MCP) tool server and an agent skill, whether as a standalone agent or inside an existing agent system. The MCP server exposes a custom-built colpack Python package that wraps HOOMD-blue hard-particle Monte Carlo, and the skill encodes a four-stage workflow contract. ColPackAgent can carry out the workflow interactively with human feedback, autonomously from an end-to-end prompt, or as autoresearch following a provided program file. Demonstrations cover cube particles in 3D, a binary system of disks and capsules in 2D, and the 2D hard-disk freezing transition, along with model-bench-m

What carries the argument

The agent skill that encodes a four-stage workflow contract, which directs the LLM agent through setup, planning, execution, and analysis steps using the tools exposed by the MCP server.

If this is right

  • ColPackAgent performs the full workflow for particle systems including 3D cubes, 2D binary mixtures, and hard-disk phase transitions.
  • The same framework supports interactive sessions with human input, single-prompt autonomous runs, and autoresearch driven by a supplied program file.
  • Stage-specific benchmarking across multiple LLMs identifies which models reliably handle setup, planning, and analysis portions of the workflow.
  • The combination converts existing simulation packages into agent-compatible research tools for studies of phase behavior and self-assembly.

Where Pith is reading between the lines

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

  • The same pattern of domain package plus MCP exposure plus workflow skill could apply to other simulation domains such as soft-matter dynamics or granular flows.
  • Dedicated tooling appears necessary to shift LLM agents from planning scientific tasks to executing them end-to-end.
  • Embedding the skill inside larger multi-agent systems could support collaborative, multi-stage research projects beyond single-workflow execution.

Load-bearing premise

General-purpose LLM agents without dedicated simulation tools and workflow instructions will only describe colloidal packing workflows rather than execute them reliably.

What would settle it

A direct test in which a general-purpose LLM receives only a high-level prompt to run a colloidal packing simulation without access to the MCP server or the agent skill and is observed to either complete the full sequence of tool calls or produce only a textual description of the steps.

Figures

Figures reproduced from arXiv: 2605.15625 by Changwoo Do, Lijie Ding.

Figure 2
Figure 2. Figure 2: FIG. 2. Overview of the colloidal packing simulation and ar [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Types of colloidal particles and the simulation pa [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Communicate with ColPackAgent interactively. (a) [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. ColPackAgent running on three MCP-and-skill [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. ColPackAgent running in autoresearch mode on the [PITH_FULL_IMAGE:figures/full_fig_p005_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. Evaluation of different LLMs on the ColPackAgent [PITH_FULL_IMAGE:figures/full_fig_p006_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: FIG. 9. The four research-output plots produced by ColPackAgent at the end of the autoresearch run reported in Sec. IV. [PITH_FULL_IMAGE:figures/full_fig_p015_9.png] view at source ↗
read the original abstract

We introduce ColPackAgent, an agent framework that autonomously runs Monte Carlo simulations of colloidal packing through a Model Context Protocol (MCP) tool server and an agent skill, whether as a standalone agent or inside an existing agent system. By harnessing the MCP server and agent skill, ColPackAgent executes a structured workflow for colloidal packing simulations, which are central to studies of phase behavior, self-assembly, and materials design. Without dedicated simulation tools and workflow instructions, general-purpose Large Language Model (LLM) agents tend to describe such workflows rather than execute them reliably. The MCP server exposes a custom-built colpack Python package that wraps HOOMD-blue hard-particle Monte Carlo, and the skill encodes a four-stage workflow contract. ColPackAgent can carry out the workflow interactively with human feedback, autonomously from an end-to-end prompt, or as autoresearch following a provided program file. We demonstrate the system in different modes with several colloidal packing simulation examples such as cube particles in 3D, a binary system of disks and capsules in 2D, and the 2D hard-disk freezing transition using autoresearch. We also compare model performance on this workflow across a panel of LLMs with 17 stage-specific prompts. This benchmark provides a stage-level check of how reliably different models follow the setup, planning, and analysis workflow. Together, these results show that pairing a domain Python package with MCP tools and a portable agent skill provides a practical route for turning a simulation toolkit into an agent-assisted research workflow.

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

2 major / 2 minor

Summary. The manuscript introduces ColPackAgent, a framework that pairs a Model Context Protocol (MCP) tool server exposing a custom colpack Python package (wrapping HOOMD-blue hard-particle Monte Carlo) with a portable agent skill encoding a four-stage workflow contract (setup, planning, execution, analysis). It demonstrates the system in standalone, interactive, and autoresearch modes on colloidal packing examples including 3D cube particles, 2D binary disk-capsule mixtures, and the 2D hard-disk freezing transition, while also reporting an LLM benchmark using 17 stage-specific prompts to evaluate workflow adherence across models.

Significance. If the central claim holds, the work offers a concrete, extensible route for converting existing domain simulation packages into agent-accessible tools, which could reduce the expertise barrier for running Monte Carlo studies of phase behavior and self-assembly. The emphasis on a portable skill and MCP server is a strength for reproducibility and integration into larger agent systems.

major comments (2)
  1. [Abstract and results on LLM benchmark] The central claim that MCP tools plus the agent skill turn a simulation toolkit into a practical agent-assisted research workflow requires that executed runs produce physically valid outputs, not merely that the agent follows the four-stage contract. The evaluation (abstract and associated results) scores LLM performance on 17 stage-specific prompts for prompt-following but reports no quantitative success rates, error analysis, or validation against known observables such as the 2D hard-disk freezing density near 0.7 or cube packing fractions. This leaves the research-utility claim only partially supported.
  2. [Demonstration sections] The demonstrations on cube particles, binary disk-capsule systems, and hard-disk freezing are described qualitatively; without tabulated metrics (e.g., achieved packing fractions, radial distribution functions, or comparison to literature values) it is difficult to assess whether the agent-driven workflows recover established physics.
minor comments (2)
  1. [Methods or workflow description] The manuscript would benefit from an explicit statement of the success criteria used to judge whether a simulation run is 'successful' in the workflow contract.
  2. [Figures] Figure captions and axis labels for any simulation output plots should include units and direct comparison to analytic or literature benchmarks where available.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thoughtful and constructive comments, which help clarify how to better demonstrate the practical research utility of ColPackAgent. We address each major comment point by point below and indicate planned revisions to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract and results on LLM benchmark] The central claim that MCP tools plus the agent skill turn a simulation toolkit into a practical agent-assisted research workflow requires that executed runs produce physically valid outputs, not merely that the agent follows the four-stage contract. The evaluation (abstract and associated results) scores LLM performance on 17 stage-specific prompts for prompt-following but reports no quantitative success rates, error analysis, or validation against known observables such as the 2D hard-disk freezing density near 0.7 or cube packing fractions. This leaves the research-utility claim only partially supported.

    Authors: The LLM benchmark section is intentionally scoped to measure workflow adherence (setup, planning, execution, analysis) across models via the 17 stage-specific prompts, directly addressing the manuscript's point that general-purpose agents tend to describe rather than execute such workflows. We agree, however, that quantitative validation of physical outputs is needed to fully support the research-utility claim. The demonstrations do produce valid Monte Carlo results in practice, but these are not yet tabulated with observables. In the revised manuscript we will add quantitative metrics, including achieved packing fractions for the 3D cube and 2D binary disk-capsule cases, observed freezing density for the hard-disk transition (compared to the literature value near 0.7), and any available error analysis or success-rate summaries. These will appear in the results section and be referenced in the abstract. revision: yes

  2. Referee: [Demonstration sections] The demonstrations on cube particles, binary disk-capsule systems, and hard-disk freezing are described qualitatively; without tabulated metrics (e.g., achieved packing fractions, radial distribution functions, or comparison to literature values) it is difficult to assess whether the agent-driven workflows recover established physics.

    Authors: The demonstrations currently emphasize the distinct operating modes (standalone, interactive, prompt-driven, and autoresearch) with qualitative descriptions of the colloidal packing outcomes. We acknowledge that this makes it harder to verify recovery of known physics. In the revision we will augment these sections with tabulated metrics: achieved packing fractions, radial distribution functions where computed, and explicit comparisons to literature values (including the hard-disk freezing density). This will provide concrete evidence that the MCP-exposed colpack workflows, guided by the agent skill, reproduce established results. revision: yes

Circularity Check

0 steps flagged

No circularity: framework demonstration with independent benchmarks

full rationale

This is a software framework and demonstration paper with no mathematical derivations, equations, fitted parameters, or self-referential predictions. The central claim—that MCP tools plus a portable agent skill turn a simulation toolkit into an agent-assisted workflow—is supported by direct demonstrations (cube particles, binary disks/capsules, 2D hard-disk freezing) and a separate 17-prompt benchmark on stage-specific LLM adherence. These evaluations are external to the framework definition itself and do not reduce to inputs by construction. No self-citation chains, uniqueness theorems, or ansatzes are invoked as load-bearing steps. The paper is self-contained against its stated benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 2 invented entities

The paper introduces a new software framework whose central claim rests on assumptions about LLM behavior and on newly created components without external validation data in the abstract.

axioms (1)
  • domain assumption General-purpose LLM agents tend to describe simulation workflows rather than execute them reliably without dedicated tools and structured instructions.
    Explicitly stated in the abstract as the motivation for introducing MCP and the agent skill.
invented entities (2)
  • ColPackAgent no independent evidence
    purpose: Autonomous execution of colloidal packing Monte Carlo workflows
    New agent framework introduced by the paper.
  • MCP tool server exposing colpack package no independent evidence
    purpose: Provide domain-specific simulation capabilities to the agent
    Custom-built component described in the abstract.

pith-pipeline@v0.9.0 · 5809 in / 1369 out tokens · 51483 ms · 2026-05-20T19:24:48.785352+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

68 extracted references · 68 canonical work pages · 1 internal anchor

  1. [1]

    2024 , month = nov, howpublished =

    Introducing the. 2024 , month = nov, howpublished =

    work page 2024

  2. [3]

    2025 , howpublished =

    Equipping Agents for the Real World with. 2025 , howpublished =

    work page 2025

  3. [4]

    Reviews of Modern Physics , volume =

    de Gennes, Pierre-Gilles , title =. Reviews of Modern Physics , volume =. 1992 , doi =

    work page 1992

  4. [5]

    and Weitz, David A

    Lu, Peter J. and Weitz, David A. , title =. Annual Review of Condensed Matter Physics , volume =. 2013 , doi =

    work page 2013

  5. [6]

    Mason, T. G. and Bibette, J. and Weitz, D. A. , title =. Physical Review Letters , volume =. 1995 , doi =

    work page 1995

  6. [7]

    and Prasad, V

    Trappe, V. and Prasad, V. and Cipelletti, Luca and Segre, P. N. and Weitz, D. A. , title =. Nature , volume =. 2001 , doi =

    work page 2001

  7. [8]

    and Solomon, Michael J

    Glotzer, Sharon C. and Solomon, Michael J. , title =. Nature Materials , volume =. 2007 , doi =

    work page 2007

  8. [9]

    and Engel, Michael and Talapin, Dmitri V

    Boles, Michael A. and Engel, Michael and Talapin, Dmitri V. , title =. Chemical Reviews , volume =. 2016 , doi =

    work page 2016

  9. [10]

    , title =

    Manoharan, Vinothan N. , title =. Science , volume =. 2015 , doi =

    work page 2015

  10. [11]

    Computer physics reports , volume=

    Molecular dynamics simulation using hard particles , author=. Computer physics reports , volume=. 1989 , publisher=

    work page 1989

  11. [12]

    and Glaser, Jens and Glotzer, Sharon C

    Anderson, Joshua A. and Glaser, Jens and Glotzer, Sharon C. , title =. Computational Materials Science , volume =. 2020 , doi =

    work page 2020

  12. [13]

    and Irrgang, M

    Anderson, Joshua A. and Irrgang, M. Eric and Glotzer, Sharon C. , title =. Computer Physics Communications , volume =. 2016 , doi =

    work page 2016

  13. [14]

    , title =

    Haji-Akbari, Amir and Glotzer, Sharon C. , title =. Journal of Physics A: Mathematical and Theoretical , volume =. 2015 , doi =

    work page 2015

  14. [15]

    and Nelson, David R

    Steinhardt, Paul J. and Nelson, David R. and Ronchetti, Marco , title =. Physical Review B , volume =. 1983 , doi =

    work page 1983

  15. [16]

    , title =

    Hansen, Jean-Pierre and McDonald, Ian R. , title =. 2013 , isbn =

    work page 2013

  16. [17]

    and Harper, Eric S

    Ramasubramani, Vyas and Dice, Bradley D. and Harper, Eric S. and Spellings, Matthew P. and Anderson, Joshua A. and Glotzer, Sharon C. , title =. Computer Physics Communications , volume =. 2020 , doi =

    work page 2020

  17. [18]

    2025 , howpublished =

    Claude. 2025 , howpublished =

    work page 2025

  18. [19]

    2025 , howpublished =

    work page 2025

  19. [20]

    2025 , howpublished =

    Karpathy, Andrej , title =. 2025 , howpublished =

    work page 2025

  20. [21]

    and Aktulga, H

    Thompson, Aidan P. and Aktulga, H. Metin and Berger, Richard and Bolintineanu, Dan S. and Brown, W. Michael and Crozier, Paul S. and. Computer Physics Communications , volume =. 2022 , doi =

    work page 2022

  21. [22]

    SoftwareX , volume =

    Abraham, Mark James and Murtola, Teemu and Schulz, Roland and P. SoftwareX , volume =. 2015 , doi =

    work page 2015

  22. [23]

    The Journal of Physical Chemistry B , volume =

    Eastman, Peter and Galvelis, Raimondas and Pel. The Journal of Physical Chemistry B , volume =. 2024 , doi =

    work page 2024

  23. [24]

    and Krauth, Werner , title =

    Bernard, Etienne P. and Krauth, Werner , title =. Physical Review Letters , volume =. 2011 , doi =

    work page 2011

  24. [25]

    and Glotzer, Sharon C

    Engel, Michael and Anderson, Joshua A. and Glotzer, Sharon C. and Isobe, Masaharu and Bernard, Etienne P. and Krauth, Werner , title =. Physical Review E , volume =. 2013 , doi =

    work page 2013

  25. [26]

    and Zheng, Xiaoyu and Petschek, Rolfe G

    Haji-Akbari, Amir and Engel, Michael and Keys, Aaron S. and Zheng, Xiaoyu and Petschek, Rolfe G. and Palffy-Muhoray, Peter and Glotzer, Sharon C. , title =. Nature , volume =. 2009 , doi =

    work page 2009

  26. [27]

    , title =

    Agarwal, Umang and Escobedo, Fernando A. , title =. Nature Materials , volume =. 2011 , doi =

    work page 2011

  27. [28]

    and Engel, Michael and Glotzer, Sharon C

    Damasceno, Pablo F. and Engel, Michael and Glotzer, Sharon C. , title =. Science , volume =. 2012 , doi =

    work page 2012

  28. [29]

    and Cox, Sam and Schilter, Oliver and Baldassari, Carlo and White, Andrew D

    Bran, Andres M. and Cox, Sam and Schilter, Oliver and Baldassari, Carlo and White, Andrew D. and Schwaller, Philippe , title =. Nature Machine Intelligence , volume =. 2024 , doi =

    work page 2024

  29. [30]

    and MacKnight, Robert and Kline, Ben and Gomes, Gabe , title =

    Boiko, Daniil A. and MacKnight, Robert and Kline, Ben and Gomes, Gabe , title =. Nature , volume =. 2023 , doi =

    work page 2023

  30. [31]

    and Rendy, Bernardus and Fei, Yuxing and Kumar, Rishi E

    Szymanski, Nathan J. and Rendy, Bernardus and Fei, Yuxing and Kumar, Rishi E. and He, Tanjin and Milsted, David and McDermott, Matthew J. and Gallant, Max and Cubuk, Ekin Dogus and Merchant, Amil and Kim, Haegyeom and Jain, Anubhav and Bartel, Christopher J. and Persson, Kristin and Zeng, Yan and Ceder, Gerbrand , title =. Nature , volume =. 2023 , doi =

    work page 2023

  31. [32]

    Journal of Applied Crystallography , volume =

    Ding, Lijie and Do, Changwoo , title =. Journal of Applied Crystallography , volume =. 2026 , doi =

    work page 2026

  32. [33]

    and Do, Changwoo , title =

    Ding, Lijie and Carrillo, Jan Michael Y. and Do, Changwoo , title =. Digital Discovery , volume =. 2026 , doi =

    work page 2026

  33. [34]

    \ de Gennes ,\ title title Soft matter , \ https://doi.org/10.1103/RevModPhys.64.645 journal journal Reviews of Modern Physics \ volume 64 ,\ pages 645--648 ( year 1992 ) NoStop

    author author P.-G. \ de Gennes ,\ title title Soft matter , \ https://doi.org/10.1103/RevModPhys.64.645 journal journal Reviews of Modern Physics \ volume 64 ,\ pages 645--648 ( year 1992 ) NoStop

    work page doi:10.1103/revmodphys.64.645 1992

  34. [35]

    author author P. J. \ Lu \ and\ author D. A. \ Weitz ,\ title title Colloidal particles: Crystals, glasses, and gels , \ https://doi.org/10.1146/annurev-conmatphys-030212-184213 journal journal Annual Review of Condensed Matter Physics \ volume 4 ,\ pages 217--233 ( year 2013 ) NoStop

    work page doi:10.1146/annurev-conmatphys-030212-184213 2013

  35. [36]

    author author T. G. \ Mason , author J. Bibette ,\ and\ author D. A. \ Weitz ,\ title title Elasticity of compressed emulsions , \ https://doi.org/10.1103/PhysRevLett.75.2051 journal journal Physical Review Letters \ volume 75 ,\ pages 2051--2054 ( year 1995 ) NoStop

    work page doi:10.1103/physrevlett.75.2051 2051

  36. [37]

    Trappe , author V

    author author V. Trappe , author V. Prasad , author L. Cipelletti , author P. N. \ Segre ,\ and\ author D. A. \ Weitz ,\ title title Jamming phase diagram for attractive particles , \ https://doi.org/10.1038/35081021 journal journal Nature \ volume 411 ,\ pages 772--775 ( year 2001 ) NoStop

    work page doi:10.1038/35081021 2001

  37. [38]

    author author S. C. \ Glotzer \ and\ author M. J. \ Solomon ,\ title title Anisotropy of building blocks and their assembly into complex structures , \ https://doi.org/10.1038/nmat1949 journal journal Nature Materials \ volume 6 ,\ pages 557--562 ( year 2007 ) NoStop

    work page doi:10.1038/nmat1949 2007

  38. [39]

    author author V. N. \ Manoharan ,\ title title Colloidal matter: Packing, geometry, and entropy , \ https://doi.org/10.1126/science.1253751 journal journal Science \ volume 349 ,\ pages 1253751 ( year 2015 ) NoStop

    work page doi:10.1126/science.1253751 2015

  39. [40]

    author author P. F. \ Damasceno , author M. Engel ,\ and\ author S. C. \ Glotzer ,\ title title Predictive self-assembly of polyhedra into complex structures , \ https://doi.org/10.1126/science.1220869 journal journal Science \ volume 337 ,\ pages 453--457 ( year 2012 ) NoStop

    work page doi:10.1126/science.1220869 2012

  40. [41]

    author author M. A. \ Boles , author M. Engel ,\ and\ author D. V. \ Talapin ,\ title title Self-Assembly of colloidal nanocrystals: From intricate structures to functional materials , \ https://doi.org/10.1021/acs.chemrev.6b00196 journal journal Chemical Reviews \ volume 116 ,\ pages 11220--11289 ( year 2016 ) NoStop

    work page doi:10.1021/acs.chemrev.6b00196 2016

  41. [42]

    author author J. A. \ Anderson , author M. E. \ Irrgang ,\ and\ author S. C. \ Glotzer ,\ title title Scalable M etropolis M onte C arlo for simulation of hard shapes , \ https://doi.org/10.1016/j.cpc.2016.02.024 journal journal Computer Physics Communications \ volume 204 ,\ pages 21--30 ( year 2016 ) NoStop

    work page doi:10.1016/j.cpc.2016.02.024 2016

  42. [43]

    \ Hansen \ and\ author I

    author author J.-P. \ Hansen \ and\ author I. R. \ McDonald ,\ @noop title Theory of Simple Liquids: with Applications to Soft Matter ,\ edition 4th \ ed.\ ( publisher Academic Press ,\ year 2013 ) NoStop

    work page 2013

  43. [44]

    author author P. J. \ Steinhardt , author D. R. \ Nelson ,\ and\ author M. Ronchetti ,\ title title Bond-orientational order in liquids and glasses , \ https://doi.org/10.1103/PhysRevB.28.784 journal journal Physical Review B \ volume 28 ,\ pages 784--805 ( year 1983 ) NoStop

    work page doi:10.1103/physrevb.28.784 1983

  44. [45]

    Haji-Akbari \ and\ author S

    author author A. Haji-Akbari \ and\ author S. C. \ Glotzer ,\ title title Strong orientational coordinates and orientational order parameters for symmetric objects , \ https://doi.org/10.1088/1751-8113/48/48/485201 journal journal Journal of Physics A: Mathematical and Theoretical \ volume 48 ,\ pages 485201 ( year 2015 ) NoStop

    work page doi:10.1088/1751-8113/48/48/485201 2015

  45. [46]

    author author E. P. \ Bernard \ and\ author W. Krauth ,\ title title Two-step melting in two dimensions: First-order liquid-hexatic transition , \ https://doi.org/10.1103/PhysRevLett.107.155704 journal journal Physical Review Letters \ volume 107 ,\ pages 155704 ( year 2011 ) NoStop

    work page doi:10.1103/physrevlett.107.155704 2011

  46. [47]

    Engel , author J

    author author M. Engel , author J. A. \ Anderson , author S. C. \ Glotzer , author M. Isobe , author E. P. \ Bernard ,\ and\ author W. Krauth ,\ title title Hard-disk equation of state: First-order liquid-hexatic transition in two dimensions with three simulation methods , \ https://doi.org/10.1103/PhysRevE.87.042134 journal journal Physical Review E \ vo...

    work page doi:10.1103/physreve.87.042134 2013

  47. [48]

    Haji-Akbari , author M

    author author A. Haji-Akbari , author M. Engel , author A. S. \ Keys , author X. Zheng , author R. G. \ Petschek , author P. Palffy-Muhoray ,\ and\ author S. C. \ Glotzer ,\ title title Disordered, quasicrystalline and crystalline phases of densely packed tetrahedra , \ https://doi.org/10.1038/nature08641 journal journal Nature \ volume 462 ,\ pages 773--...

    work page doi:10.1038/nature08641 2009

  48. [49]

    Agarwal \ and\ author F

    author author U. Agarwal \ and\ author F. A. \ Escobedo ,\ title title Mesophase behaviour of polyhedral particles , \ https://doi.org/10.1038/nmat2959 journal journal Nature Materials \ volume 10 ,\ pages 230--235 ( year 2011 ) NoStop

    work page doi:10.1038/nmat2959 2011

  49. [50]

    Karpathy ,\ @noop title autoresearch , \ howpublished https://github.com/karpathy/autoresearch ( year 2025 ) NoStop

    author author A. Karpathy ,\ @noop title autoresearch , \ howpublished https://github.com/karpathy/autoresearch ( year 2025 ) NoStop

    work page 2025

  50. [51]

    author author A. M. \ Bran , author S. Cox , author O. Schilter , author C. Baldassari , author A. D. \ White ,\ and\ author P. Schwaller ,\ title title Augmenting large language models with chemistry tools , \ https://doi.org/10.1038/s42256-024-00832-8 journal journal Nature Machine Intelligence \ volume 6 ,\ pages 525--535 ( year 2024 ) NoStop

    work page doi:10.1038/s42256-024-00832-8 2024

  51. [52]

    author author D. A. \ Boiko , author R. MacKnight , author B. Kline ,\ and\ author G. Gomes ,\ title title Autonomous chemical research with large language models , \ https://doi.org/10.1038/s41586-023-06792-0 journal journal Nature \ volume 624 ,\ pages 570--578 ( year 2023 ) NoStop

    work page doi:10.1038/s41586-023-06792-0 2023

  52. [53]

    author author N. J. \ Szymanski , author B. Rendy , author Y. Fei , author R. E. \ Kumar , author T. He , author D. Milsted , author M. J. \ McDermott , author M. Gallant , author E. D. \ Cubuk , author A. Merchant , author H. Kim , author A. Jain , author C. J. \ Bartel , author K. Persson , author Y. Zeng ,\ and\ author G. Ceder ,\ title title An autono...

    work page doi:10.1038/s41586-023-06734-w 2023

  53. [54]

    Ding \ and\ author C

    author author L. Ding \ and\ author C. Do ,\ title title SasAgent : multi-agent artificial intelligence system for small-angle scattering data analysis , \ https://doi.org/10.1107/S1600576726001299 journal journal Journal of Applied Crystallography \ volume 59 ,\ pages 578 ( year 2026 ) NoStop

    work page doi:10.1107/s1600576726001299 2026

  54. [55]

    Ding , author J

    author author L. Ding , author J. M. Y. \ Carrillo ,\ and\ author C. Do ,\ title title ToPolyAgent : AI agents for coarse-grained bead-spring topological polymer simulations , \ https://doi.org/10.1039/D5DD00471C journal journal Digital Discovery \ volume 5 ,\ pages 901--909 ( year 2026 ) NoStop

    work page doi:10.1039/d5dd00471c 2026

  55. [56]

    author author Anthropic ,\ @noop title Introducing the Model Context Protocol , \ howpublished https://www.anthropic.com/news/model-context-protocol ( year 2024 ) NoStop

    work page 2024

  56. [57]

    Model Context Protocol (MCP): Landscape, Security Threats, and Future Research Directions

    author author X. Hou , author Y. Zhao , author S. Wang ,\ and\ author H. Wang ,\ title title Model context protocol ( MCP ): Landscape, security threats, and future research directions , \ https://doi.org/10.48550/arXiv.2503.23278 journal journal arXiv preprint \ ( year 2025 ),\ 10.48550/arXiv.2503.23278 ,\ https://arxiv.org/abs/2503.23278 2503.23278 NoStop

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2503.23278 2025

  57. [58]

    author author Anthropic ,\ @noop title Equipping agents for the real world with Agent Skills , \ howpublished https://www.anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills ( year 2025 a ) NoStop

    work page 2025

  58. [59]

    author author J. A. \ Anderson , author J. Glaser ,\ and\ author S. C. \ Glotzer ,\ title title HOOMD-blue : A P ython package for high-performance molecular dynamics and hard particle M onte C arlo simulations , \ https://doi.org/10.1016/j.commatsci.2019.109363 journal journal Computational Materials Science \ volume 173 ,\ pages 109363 ( year 2020 ) NoStop

    work page doi:10.1016/j.commatsci.2019.109363 2019

  59. [60]

    Ramasubramani , author B

    author author V. Ramasubramani , author B. D. \ Dice , author E. S. \ Harper , author M. P. \ Spellings , author J. A. \ Anderson ,\ and\ author S. C. \ Glotzer ,\ title title freud: A software suite for high throughput analysis of particle simulation data , \ https://doi.org/10.1016/j.cpc.2020.107275 journal journal Computer Physics Communications \ volu...

    work page doi:10.1016/j.cpc.2020.107275 2020

  60. [61]

    author author M. P. \ Allen , author D. Frenkel ,\ and\ author J. Talbot ,\ title title Molecular dynamics simulation using hard particles , \ @noop journal journal Computer physics reports \ volume 9 ,\ pages 301--353 ( year 1989 ) NoStop

    work page 1989

  61. [62]

    author author Anthropic ,\ @noop title Claude C ode: An agentic coding tool that lives in your terminal , \ howpublished https://github.com/anthropics/claude-code ( year 2025 b ) NoStop

    work page 2025

  62. [63]

    author author Google ,\ @noop title Gemini CLI : An open-source AI agent that brings the power of Gemini directly into your terminal , \ howpublished https://github.com/google-gemini/gemini-cli ( year 2025 ) NoStop

    work page 2025

  63. [64]

    author author OpenAI ,\ @noop title Codex : Lightweight coding agent that runs in your terminal , \ howpublished https://github.com/openai/codex ( year 2025 ) NoStop

    work page 2025

  64. [65]

    author author Anomaly ,\ @noop title OpenCode : The open source coding agent , \ howpublished https://github.com/anomalyco/opencode ( year 2025 ) NoStop

    work page 2025

  65. [66]

    author author OpenRouter ,\ @noop title OpenRouter : A unified interface for LLM s , \ howpublished https://openrouter.ai/ ( year 2025 ) NoStop

    work page 2025

  66. [67]

    author author A. P. \ Thompson , author H. M. \ Aktulga , author R. Berger , author D. S. \ Bolintineanu , author W. M. \ Brown , author P. S. \ Crozier , author P. J. \ in 't Veld , author A. Kohlmeyer , author S. G. \ Moore , author T. D. \ Nguyen , author R. Shan , author M. J. \ Stevens , author J. Tranchida , author C. Trott ,\ and\ author S. J. \ Pl...

    work page doi:10.1016/j.cpc.2021.108171 2021

  67. [68]

    author author M. J. \ Abraham , author T. Murtola , author R. Schulz , author S. P \'a ll , author J. C. \ Smith , author B. Hess ,\ and\ author E. Lindahl ,\ title title GROMACS : High performance molecular simulations through multi-level parallelism from laptops to supercomputers , \ https://doi.org/10.1016/j.softx.2015.06.001 journal journal SoftwareX ...

    work page doi:10.1016/j.softx.2015.06.001 2015

  68. [69]

    Eastman , author R

    author author P. Eastman , author R. Galvelis , author R. P. \ Pel \'a ez , author C. R. A. \ Abreu , author S. E. \ Farr , author E. Gallicchio , author A. Gorenko , author M. M. \ Henry , author F. Hu , author J. Huang , author A. Kr \"a mer , author J. Michel , author J. A. \ Mitchell , author V. S. \ Pande , author J. P. G. L. M. \ Rodrigues , author ...

    work page doi:10.1021/acs.jpcb.3c06662 2024