pith. sign in

arxiv: 2605.15227 · v1 · pith:BCS3MHD7new · submitted 2026-05-13 · 💻 cs.AI · cond-mat.mtrl-sci· cs.RO

NIMO Controller: a self-driving laboratory orchestrator based on the Model Context Protocol

Naruki Yoshikawa , Ryo Tamura This is my paper

Pith reviewed 2026-05-19 17:42 UTC · model grok-4.3

classification 💻 cs.AI cond-mat.mtrl-scics.RO
keywords self-driving laboratoriesModel Context ProtocolSDL orchestratorAI agentsvisual programming interfaceexperimental workflowsNIMO Controller
0
0 comments X

The pith

MCP-based architecture unifies interfaces for humans and AI agents in self-driving labs

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

The paper proposes an SDL software architecture built around the Model Context Protocol, where every laboratory functionality is exposed as an MCP server. This single backend automatically generates a visual programming interface so human users can assemble experimental workflows without writing code. The identical MCP servers also allow AI agents to access and control the same components directly. A case study on a color-matching self-driving laboratory confirms that the NIMO Controller can orchestrate real hardware and software under this design. A sympathetic reader would care because the approach aims to lower the technical barriers that currently separate human operators from AI-driven scientific automation.

Core claim

We propose an SDL software architecture based on the Model Context Protocol (MCP), in which all SDL functionalities are exposed through MCP servers. Following this design principle, we introduce an MCP-based SDL orchestrator, named NIMO Controller. It provides a visual programming interface automatically generated through MCP-based tool discovery, allowing human users to design experimental workflows without writing code. The same MCP backend can also be accessed by AI agents, providing a unified interface for both human users and AI agents.

What carries the argument

Model Context Protocol (MCP) servers that expose all SDL functionalities, enabling automatic tool discovery for visual interfaces and direct access by AI agents.

If this is right

  • Human users can design and execute experimental workflows using a visual interface generated automatically from MCP tool discovery.
  • AI agents control the same laboratory components through the identical MCP backend used by human operators.
  • The architecture coordinates real laboratory hardware and software via standardized MCP interfaces rather than bespoke integrations.
  • The approach is validated by a working case study on a color-matching self-driving laboratory.

Where Pith is reading between the lines

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

  • This unified MCP layer could let AI agents autonomously propose, modify, and execute full experimental campaigns inside self-driving labs.
  • Widespread use of MCP servers might reduce the custom software glue needed when connecting new instruments across different SDL installations.
  • The same server-exposure pattern could be applied to other automated scientific platforms such as robotic synthesis systems or high-throughput screening setups.

Load-bearing premise

That exposing all SDL functionalities through MCP servers is technically feasible and sufficient to coordinate real laboratory hardware and software components without major custom integration work.

What would settle it

A demonstration in which an AI agent uses only the NIMO Controller's MCP interface to run a new physical experiment and successfully completes the workflow without any additional custom coding or hardware-specific adapters.

Figures

Figures reproduced from arXiv: 2605.15227 by Naruki Yoshikawa, Ryo Tamura.

Figure 1
Figure 1. Figure 1: Architecture of NIMO Controller. The controller serves as an MCP host that integrates [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Hardware setup of the color-matching SDL. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: A screenshot of NIMO Controller. The left side shows the experimental workflow designed [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Results of the color-matching SDL experiments. Top row: photographs of the 12-well [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
read the original abstract

Self-driving laboratories (SDLs) have attracted increasing attention as a means of accelerating scientific discovery; however, developing SDL software remains technically demanding. To improve accessibility, orchestration software frameworks have been proposed to coordinate SDL components. Nevertheless, existing frameworks are primarily designed for human interaction and do not provide standardized interfaces suitable for AI agents. In this work, we propose an SDL software architecture based on the Model Context Protocol (MCP), in which all SDL functionalities are exposed through MCP servers. Following this design principle, we introduce an MCP-based SDL orchestrator, named NIMO Controller. It provides a visual programming interface automatically generated through MCP-based tool discovery, allowing human users to design experimental workflows without writing code. The same MCP backend can also be accessed by AI agents, providing a unified interface for both human users and AI agents. We demonstrate the proposed system through a case study on a color-matching SDL. The results validate the usability of the proposed MCP-based SDL architecture.

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 / 1 minor

Summary. The manuscript proposes an SDL software architecture based on the Model Context Protocol (MCP) and introduces the NIMO Controller as an orchestrator. All SDL functionalities are exposed through MCP servers, enabling an automatically generated visual programming interface for human users via tool discovery and direct access by AI agents through the same backend. This is presented as providing a unified interface that improves accessibility without requiring major custom integration work. The approach is demonstrated and claimed to be validated through a case study implementing a color-matching self-driving laboratory.

Significance. If the architecture can be shown to coordinate real laboratory hardware and software components reliably, the work could advance standardization in SDL orchestration by bridging human and AI interfaces through a single MCP-based backend. The concrete implementation of NIMO Controller together with the color-matching case study demonstration is a strength, as it provides a practical example of the proposed design principle rather than remaining purely conceptual.

major comments (1)
  1. [Case Study] Case Study section: The manuscript states that the color-matching SDL case study validates the usability of the MCP-based architecture, yet no quantitative performance metrics (e.g., workflow completion rates, latency measurements, or error rates), user studies, or comparisons to existing orchestrators are reported. This is load-bearing for the central claim of improved accessibility and feasibility without major custom integration, as the abstract and introduction position the demonstration as evidence supporting these benefits.
minor comments (1)
  1. [Abstract] The abstract refers to 'the results' validating usability without specifying what those results consist of; adding a brief indication of the evaluation criteria used in the case study would improve clarity.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback and for recognizing the practical value of the NIMO Controller implementation together with the color-matching demonstration. We address the single major comment below and have prepared a revised manuscript that incorporates additional details from the existing case study data.

read point-by-point responses

  1. Referee: The manuscript states that the color-matching SDL case study validates the usability of the MCP-based architecture, yet no quantitative performance metrics (e.g., workflow completion rates, latency measurements, or error rates), user studies, or comparisons to existing orchestrators are reported. This is load-bearing for the central claim of improved accessibility and feasibility without major custom integration, as the abstract and introduction position the demonstration as evidence supporting these benefits.

    Authors: We agree that quantitative metrics would provide stronger support for the claims of accessibility and reduced integration effort. The color-matching case study was executed as an end-to-end demonstration on physical hardware, and we can extract additional performance data (such as average latency per MCP tool invocation and observed workflow success rate across the ten runs performed) directly from the logs already collected during the experiments. In the revised manuscript we will add a new subsection under the case study reporting these measurements along with a short discussion of any transient errors encountered. We will also expand the related-work section with a concise comparison table contrasting the unified MCP backend against the human-centric interfaces of ChemOS, ARES, and similar frameworks, emphasizing the absence of custom API wrappers required for AI-agent access. User studies fall outside the scope of the current technical contribution, which centers on architecture and implementation feasibility rather than human-factors evaluation; we will explicitly state this limitation and note it as future work. revision: partial

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper proposes an MCP-based SDL software architecture and NIMO Controller orchestrator, validated through a color-matching case study implementation. No derivation chain, equations, fitted parameters presented as predictions, or load-bearing self-citations exist; the central claims rest on the feasibility and usability shown in the independent demonstration rather than reducing to inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim rests on the domain assumption that MCP servers can expose and coordinate all SDL components, plus the newly introduced NIMO Controller entity; no free parameters or fitted values are described.

axioms (1)
  • domain assumption All SDL functionalities can be exposed through MCP servers to provide standardized interfaces suitable for both humans and AI agents
    This premise is invoked in the abstract when stating the proposed architecture and the unified interface.
invented entities (1)
  • NIMO Controller no independent evidence
    purpose: MCP-based SDL orchestrator that auto-generates visual programming interfaces
    New system introduced by the authors to implement the MCP architecture.

pith-pipeline@v0.9.0 · 5703 in / 1255 out tokens · 41011 ms · 2026-05-19T17:42:09.028798+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

13 extracted references · 13 canonical work pages

  1. [1]

    Self-Driving Laboratories for Chemistry and Materials Science.Chemical Reviews, 2024

    Gary Tom, Stefan P. Schmid, Sterling G. Baird, Yang Cao, Kourosh Darvish, Han Hao, Stanley Lo, Sergio Pablo-Garc´ ıa, Ella M. Rajaonson, Marta Skreta, Naruki Yoshikawa, Samantha Corapi, Gun Deniz Akkoc, Felix Strieth-Kalthoff, Martin Seifrid, and Al´ an Aspuru-Guzik. Self-driving laboratories for chemistry and materials science.Chemical Reviews, 124(16):9...

    work page doi:10.1021/acs.chemrev.4c00055 2024

  2. [2]

    Chemos: An orchestratio n software to democratize autonomous discovery

    Lo¨ ıc M. Roch, Florian H¨ ase, Christoph Kreisbeck, Teresa Tamayo-Mendoza, Lars P. E. Yunker, Jason E. Hein, and Al´ an Aspuru-Guzik. ChemOS: An orchestration software to democratize autonomous discovery.PLOS ONE, 15(4):1–18, 04 2020. doi: 10.1371/journal.pone.0229862

    work page doi:10.1371/journal.pone.0229862 2020

  3. [3]

    Hickman, Santiago Miret, Sergio Pablo-Garc´ ıa, and Al´ an Aspuru-Guzik

    Malcolm Sim, Mohammad Ghazi Vakili, Felix Strieth-Kalthoff, Han Hao, Riley J. Hickman, Santiago Miret, Sergio Pablo-Garc´ ıa, and Al´ an Aspuru-Guzik. ChemOS 2.0: An orchestration architecture for chemical self-driving laboratories.Matter, 7(9):2959–2977, 2024. doi: 10.1016/ j.matt.2024.04.022

    work page 2024

  4. [4]

    Sahasrabuddhe, Matthew J

    Yuxing Fei, Bernardus Rendy, Rishi Kumar, Olympia Dartsi, Hrushikesh P. Sahasrabuddhe, Matthew J. McDermott, Zheren Wang, Nathan J. Szymanski, Lauren N. Walters, David Mil- sted, Yan Zeng, Anubhav Jain, and Gerbrand Ceder. AlabOS: a Python-based reconfigurable workflow management framework for autonomous laboratories.Digital Discovery, 3(11):2275– 2288, 2...

    work page doi:10.1039/d4dd00129j 2024

  5. [5]

    Reish, Rebekah Greenwood, Noah Depner, Jiyoon Min, Rama El-khawaldeh, Paloma Prieto, Ekaterina Trushina, and Jason E

    Wenyu Zhang, Lucy Hao, Veronica Lai, Ryan Corkery, Jacob Jessiman, Jiayu Zhang, Junliang Liu, Yusuke Sato, Maria Politi, Matthew E. Reish, Rebekah Greenwood, Noah Depner, Jiyoon Min, Rama El-khawaldeh, Paloma Prieto, Ekaterina Trushina, and Jason E. Hein. IvoryOS: an interoperable web interface for orchestrating Python-based self-driving laboratories.Natu...

    work page doi:10.1038/s41467-025-60514-w 2025

  6. [6]

    Ryo Tamura, Koji Tsuda, and Shoichi Matsuda. NIMS-OS: an automation software to im- plement a closed loop between artificial intelligence and robotic experiments in materials sci- ence.Science and Technology of Advanced Materials: Methods, 3(1):2232297, 2023. doi: 10.1080/27660400.2023.2232297

    work page doi:10.1080/27660400.2023.2232297 2023

  7. [7]

    Bridging decision-making engines and workflow design in self-driving laboratories: A NIMO–IvoryOS integration study.ChemRxiv, 2026

    Naruki Yoshikawa, Taiga Ozawa, Wenyu Zhang, Jason Hein, Ryo Tamura, and Matsuda Shoichi. Bridging decision-making engines and workflow design in self-driving laboratories: A NIMO–IvoryOS integration study.ChemRxiv, 2026. doi: 10.26434/chemrxiv.15000670/v1

    work page doi:10.26434/chemrxiv.15000670/v1 2026

  8. [8]

    Ryo Tamura, Hiromichi Taketa, Satoshi Murata, Daisuke Ryuno, Tomotaka Yokota, Koji Tsuda, and Shoichi Matsuda. Seamless integration of legacy robotic systems into a self-driving laboratory via NIMO: a case study on liquid handler automation.Science and Technology of Advanced Materials: Methods, 5(1):2565144, 2025. doi: 10.1080/27660400.2025.2565144

    work page doi:10.1080/27660400.2025.2565144 2025

  9. [9]

    https://modelcontextprotocol.io/specification/2025- 11-25, 2025

    Specification - model context protocol. https://modelcontextprotocol.io/specification/2025- 11-25, 2025. Accessed: May 9, 2026

    work page 2025

  10. [10]

    What is a minimal working example for a self-driving laboratory?Matter, 5(12):4170–4178, 2022

    Sterling G Baird and Taylor D Sparks. What is a minimal working example for a self-driving laboratory?Matter, 5(12):4170–4178, 2022. doi: 10.1016/j.matt.2022.11.007. 8

    work page doi:10.1016/j.matt.2022.11.007 2022

  11. [11]

    Gaurav Sharma, Wencheng Wu, and Edul N. Dalal. The CIEDE2000 color-difference for- mula: Implementation notes, supplementary test data, and mathematical observations.Color Research & Application, 30(1):21–30, 2004. doi: 10.1002/col.20070

    work page doi:10.1002/col.20070 2004

  12. [12]

    Bayesian optimization package: PHYSBO.Computer Physics Communications, 278: 108405, 2022

    Yuichi Motoyama, Ryo Tamura, Kazuyoshi Yoshimi, Kei Terayama, Tsuyoshi Ueno, and Koji Tsuda. Bayesian optimization package: PHYSBO.Computer Physics Communications, 278: 108405, 2022. doi: 10.1016/j.cpc.2022.108405

    work page doi:10.1016/j.cpc.2022.108405 2022

  13. [13]

    Update of PHYSBO: Improving usability and portability of bayesian optimization for physics and materials research.arXiv preprint, 2026

    Yuichi Motoyama, Kazuyoshi Yoshimi, Tatsumi Aoyama, Kei Terayama, Koji Tsuda, and Ryo Tamura. Update of PHYSBO: Improving usability and portability of bayesian optimization for physics and materials research.arXiv preprint, 2026. doi: 10.48550/arXiv.2603.01349. 9

    work page doi:10.48550/arxiv.2603.01349 2026