Tool-Surface Granularity in Model Context Protocol Servers for Analytical Engines: Patterns, Trade-Offs, and Token Economics
Keywords:
Model Context Protocol, large language models, agentic systems, function calling, ReAct, tool design, token economics, analytical enginesAbstract
The Model Context Protocol (MCP) has emerged as the standard interface between large language models and external analytical systems. The design discipline governing the tool surface that an MCP server exposes remains under-articulated in the literature and underspecified in production deployments. The dominant pattern in early deployments is a single monolithic query tool accepting an arbitrary SQL string; the principled alternative is fine-grained partitioning into one tool per logical capability, defined as the pair (verb × resource-type). This article argues that the choice between these patterns is a multi-objective design decision in which token economics is one dimension among several. A parameterized cost model for Reason-and-Act agent loops with function-calling identifies the parameter regimes in which each pattern dominates. A 30-task synthetic token-cost calculator shows that the monolithic pattern is token-cheaper across the modeled range under a conservative reasoning multiplier of μ = 4, while the fine-grained pattern becomes token-competitive at task lengths of approximately four to six calls when dialect-translation cost is factored in. Capability-scoped authorization, replayable audit logs, and reasoning-chain shortening under high dialect-translation cost justify the fine-grained pattern's token premium in production environments, independent of any token-cost cross-over. A protocol-level design discipline for analytical MCP servers and a parameterized cost framework that operators can apply to their own workload distributions are provided as concrete contributions.
Downloads
References
J. H. Saltzer and M. D. Schroeder, "The Protection of Information in Computer Systems," Proceedings of the IEEE, vol. 63, no. 9, pp. 1278–1308, 1975. DOI: 10.1109/PROC.1975.9939. Available: https://ieeexplore.ieee.org/document/1451869
M. S. Miller, "Robust Composition: Towards a Unified Approach to Access Control and Concurrency Control," Ph.D. dissertation, Johns Hopkins University, 2006. Available: http://www.erights.org/talks/thesis/
Anthropic, "Model Context Protocol Specification," 2024. Available: https://modelcontextprotocol.io/specification/2025-11-25
S. Yao, J. Zhao, D. Yu, N. Du, I. Shafran, K. Narasimhan, and Y. Cao, "ReAct: Synergizing Reasoning and Acting in Language Models," Proc. ICLR, 2023. Available: https://arxiv.org/abs/2210.03629
OpenAI, "Function Calling in the OpenAI API," OpenAI Documentation, 2024. Available: https://platform.openai.com/docs/guides/function-calling
Anthropic, "Tool Use with Claude," Anthropic Documentation, 2024. Available: https://docs.anthropic.com/en/docs/build-with-claude/tool-use
S. G. Patil, T. Zhang, X. Wang, and J. E. Gonzalez, "Gorilla: Large Language Model Connected with Massive APIs," Advances in NeurIPS, vol. 37, 2024. Available: https://arxiv.org/abs/2305.15334
X. Liu et al., "AgentBench: Evaluating LLMs as Agents," Proc. ICLR, 2024. Available: https://arxiv.org/abs/2308.03688
T. Schick, J. Dwivedi-Yu, R. Dessì, R. Raileanu, M. Lomeli, E. Hambro, L. Zettlemoyer, N. Cancedda, and T. Scialom, "Toolformer: Language Models Can Teach Themselves to Use Tools," Advances in NeurIPS, vol. 36, 2023. Available: https://arxiv.org/abs/2302.04761
Y. Qin et al., "ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs," Proc. ICLR (Spotlight), 2024. Available: https://arxiv.org/abs/2307.16789
J. Wei, X. Wang, D. Schuurmans, M. Bosma, B. Ichter, F. Xia, E. Chi, Q. Le, and D. Zhou, "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models," Advances in NeurIPS, vol. 35, 2022. Available: https://arxiv.org/abs/2201.11903
Z. Xi et al., "The Rise and Potential of Large Language Model Based Agents: A Survey," arXiv:2309.07864, 2023. Available: https://arxiv.org/abs/2309.07864
G. Wang et al., "Voyager: An Open-Ended Embodied Agent with Large Language Models," Advances in NeurIPS, vol. 36, 2023. Available: https://arxiv.org/abs/2305.16291
G. Mialon, R. Dessì, M. Lomeli, C. Nalmpantis, R. Pasunuru, R. Raileanu, B. Rozière, T. Schick, J. Dwivedi-Yu, A. Celikyilmaz, E. Grave, Y. LeCun, and T. Scialom, "Augmented Language Models: a Survey," Transactions on Machine Learning Research, 2023. Available: https://arxiv.org/abs/2302.07842
N. Shinn, F. Cassano, E. Berman, A. Gopinath, K. Narasimhan, and S. Yao, "Reflexion: Language Agents with Verbal Reinforcement Learning," Advances in NeurIPS, vol. 36, 2023. Available: https://arxiv.org/abs/2303.11366
A. Zeng et al., "AgentTuning: Enabling Generalized Agent Abilities for LLMs," arXiv:2310.12823, 2023. Available: https://arxiv.org/abs/2310.12823
A. Parisi, Y. Zhao, and N. Fiedel, "TALM: Tool Augmented Language Models," arXiv:2205.12255, 2022. Available: https://arxiv.org/abs/2205.12255
Q. Dong et al., "A Survey on In-Context Learning," arXiv:2301.00234, 2023. Available: https://arxiv.org/abs/2301.00234
M. Kinniment et al., "Evaluating Language-Model Agents on Realistic Autonomous Tasks," arXiv:2312.11671, 2023. Available: https://arxiv.org/abs/2312.11671
Y. Lai, C. Li, Y. Wang, T. Zhang, R. Zhong, L. Zettlemoyer, S. W. Yih, D. Fried, S. Wang, and T. Yu, "DS-1000: A Natural and Reliable Benchmark for Data Science Code Generation," Proc. ICML, 2023. Available: https://arxiv.org/abs/2211.11501
Y. Shen et al., "HuggingGPT: Solving AI Tasks with ChatGPT and Its Friends in Hugging Face," Advances in NeurIPS, vol. 36, 2023. Available: https://arxiv.org/abs/2303.17580
C. C. Ji, H. Mao, F. Yan, S. G. Patil, T. Zhang, I. Stoica, and J. E. Gonzalez, "Gorilla OpenFunctions-v2," UC Berkeley Technical Report, 2024. Available: https://gorilla.cs.berkeley.edu/blogs/7_open_functions_v2.html
Downloads
Published
How to Cite
Issue
Section
License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
All papers should be submitted electronically. All submitted manuscripts must be original work that is not under submission at another journal or under consideration for publication in another form, such as a monograph or chapter of a book. Authors of submitted papers are obligated not to submit their paper for publication elsewhere until an editorial decision is rendered on their submission. Further, authors of accepted papers are prohibited from publishing the results in other publications that appear before the paper is published in the Journal unless they receive approval for doing so from the Editor-In-Chief.
IJISAE open access articles are licensed under a Creative Commons Attribution-ShareAlike 4.0 International License. This license lets the audience to give appropriate credit, provide a link to the license, and indicate if changes were made and if they remix, transform, or build upon the material, they must distribute contributions under the same license as the original.
