A Survey of the Model Context Protocol (MCP): Standardizing Context to Enhance Large Language Models (LLMs)
Open accessAbstract
The Model Context Protocol (MCP), recently introduced by Anthropic, proposes a standardized framework for integrating large language models (LLMs) with dynamic external systems. This survey reviews the foundational architecture of MCP, including its client-server model, standardized messaging protocols, dynamic tool discovery, and security mechanisms, against the backdrop of current, fragmented API solutions. Although the protocol promises enhanced interoperability and scalability for Agentic AI systems, data supporting its long-term performance remains limited. MCP’s design is critically evaluated, potential applications in domains such as finance, healthcare, and customer service are discussed, and the key challenges are outlined. This work aims to inform researchers and practitioners of MCP’s potential benefits and its present limitations in the evolving AI integration landscape. The GitHub link for this survey is: Model-Context-Protocol-Survey.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
MCP enables real-time policy enforcement: every tool invocation or data query from an agent can be evaluated against security and privacy rules before execution, with the MCP server blocking or flagging any disallowed actions automatically.
“Crucially, MCP enables real-time policy enforcement: every tool invocation or data query from an agent can be evaluated against security and privacy rules before execution, with the MCP server blocking or flagging any disallowed actions automatically [52].”
From A Review of Multi-Agent AI Systems for Biological and Clinical Data Analysis (2026) · cites this paper as [52]
Section 4.3 (Tools; page unavailable) describes structured tool calls, input validation, and controlled execution. The Security & Validation and 4.1 (Resources) passages also mention controlled execution policies and access control. None says that every tool invocation or data query is checked before execution or that an MCP server automatically blocks or flags disallowed actions. (LLM, conf=0.84)
Experimental — accuracy not yet measuredThe proposed Model Context Protocol (MCP) mediates agent access to external data with fine-grained security rules, acting as an intelligent broker that filters queries and enforces policies.
“for instance, the proposed Model Context Protocol (MCP) mediates agent access to external data with fine-grained security rules, acting as an intelligent broker that filters queries and enforces policies [52,53,64].”
From A Review of Multi-Agent AI Systems for Biological and Clinical Data Analysis (2026) · cites this paper as [52]
The Introduction describes MCP as connecting AI models to external resources through servers that securely access data, and the Security and Security & Validation sections describe authorization and controlled execution policies (pages not provided). These passages support mediated, policy-controlled access, but they do not describe MCP as an intelligent broker that filters queries or establish the claimed fine-grained security rules. (LLM, conf=0.82)
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
The publisher hasn’t deposited this paper’s reference list.
Research works printed but not linked: 6
- Zhao, W.X.; Zhou, K.; Li, J.; Tang, T.; Wang, X.; Hou, Y.; Min, Y.; Zhang, B.; Zhang, J.; Dong, Z.; et al. A Survey of Large Language Models, 2025, [arXiv:cs.CL/2303.18223].
- Perron, B.E.; Luan, H.; Qi, Z.; Victor, B.G.; Goyal, K. Demystifying Application Programming Interfaces (APIs): Unlocking the Power of Large Language Models and Other Web-based AI Services in Social Work Research, 2024, [arXiv:cs.SE/2410.20211].
- Kaddour, J.; Harris, J.; Mozes, M.; Bradley, H.; Raileanu, R.; McHardy, R. Challenges and Applications of Large Language Models, 2023, [arXiv:cs.CL/2307.10169].
- Chen, X.; Gao, C.; Chen, C.; Zhang, G.; Liu, Y. An Empirical Study on Challenges for LLM Application Developers, 2025, [arXiv:cs.SE/2408.05002].
- White, J.; Fu, Q.; Hays, S.; Sandborn, M.; Olea, C.; Gilbert, H.; Elnashar, A.; Spencer-Smith, J.; Schmidt, D.C. A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT, 2023, [arXiv:cs.SE/2302.11382].
- Qin, Y.; Liang, S.; Ye, Y.; Zhu, K.; Yan, L.; Lu, Y.; Lin, Y.; Cong, X.; Tang, X.; Qian, B.; et al. ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs, 2023, [arXiv:cs.AI/2307.16789].
Cited by
Sorted by citations
LLM-Powered AI Agent Systems and Their Applications in Industry
Liang, G., & Tong, Q. (2025) · IEEE World AI IoT Congress (AIIoT) · doi:10.1109/aiiot65859.2025.11105299 · Cited by 32 · see citing works
Ehtesham, A., Singh, A., & Kumar, S. (2025) · IEEE World AI IoT Congress (AIIoT) · doi:10.1109/aiiot65859.2025.11105280 · Cited by 13 · see citing works
Self-citationJung, H. W., Park, J. Y., Holoubek, T., Kim, W. J., & Park, J. (2025) · International Journal of Social Robotics · doi:10.1007/s12369-025-01323-5 · Cited by 7 · see citing works
Huang, Y., Zheng, B., Dong, Y., Tang, H., Zhao, H., Shawon, R. H., An, S., & Zhang, H. (2025) · Nanophotonics · doi:10.1515/nanoph-2025-0507 · Cited by 7 · see citing works
From foundation models to autonomous agents in biology
Huang, S., Lang, M., Chen, Z., Yang, C., Huang, X., Mohtashaminia, Z., & Peng, Y. (2026) · Genomics Communications · doi:10.48130/gcomm-0026-0005 · Cited by 3 · see citing works
Model Context Contracts - MCP-Enabled Framework to Integrate LLMs With Blockchain
Bandara, E., Shetty, S. S., Mukkamala, R., Gore, R. J., Rahman, A., Foytik, P. B., Bouk, S. H., Liang, X., Ng, W. K., & Zoysa, K. D. (2025) · International Conference on Blockchain Computing and Applications (BCCA) · doi:10.1109/bcca66705.2025.11229749 · Cited by 2 · see citing works
Ma, F., Li, D., Chen, Y., Xing, J., Liu, Y., Lan, D., Shao, J., & Pang, Z. (2025) · Research · doi:10.34133/research.1047 · Cited by 3 · see citing works
Slice-MCP — Fine-Tuned Llama-4 LLM and MCP Enabled 5G Network Slice Orchestration Platform
Bandara, E., Bouk, S. H., Mukkamala, R., Rahman, A., Liang, X., Gore, R. J., & Shetty, S. S. (2025) · IEEE Military Communications Conference (MILCOM) · doi:10.1109/milcom64451.2025.11310417 · Cited by 2 · see citing works
A Large-Scale Evolvable Dataset for Model Context Protocol Ecosystem and Security Analysis
Lin, Z., Ruan, B., Liu, J., & Zhao, W. (2025) · IEEE/ACM International Conference on Automated Software Engineering (ASE) · doi:10.1109/ase63991.2025.00356 · Cited by 2 · see citing works
An Agentic AI Control Plane for 6G Network Slice Orchestration, Monitoring, and Trading
Bandara, E., Gore, R., Shetty, S. S., Mukkamala, R., Hewa, T., Rahman, A., Liang, X., Bouk, S. H., Foytik, P. B., Keong, N. W., et al. (2026) · International Wireless Communications and Mobile Computing Conference (IWCMC) · doi:10.1109/iwcmc69287.2026.11580069 · Cited by 2 · see citing works
A Review of Multi-Agent AI Systems for Biological and Clinical Data Analysis
Spieser, J., Balapour, A., Meller, J., Patra, K. C., & Shamsaei, B. (2026) · Methods and Protocols · doi:10.3390/mps9020033 · Cited by 1 · see citing works
Research on Standardized Encapsulation of Model Context Protocol Tools and AI Agent Integration
Gao, M., Li, L., Ming, X., & Shen, G. (2025) · International Conference on Aerospace Information Perception and Intelligent Processing (AIPIP) · doi:10.1109/aipip66876.2025.11299220 · Cited by 2 · see citing works
Ho, T.-L., & Lam, T.-P. (2026) · Applied Sciences · doi:10.3390/app16094400 · Cited by 1 · see citing works
Enhancing Digital Product Passports for the Circular Economy with Generative AI
Pourjafarian, M., Plociennik, C., Stein, P., Moarefvand, N., & Ruskowski, M. (2026) · Procedia CIRP · doi:10.1016/j.procir.2026.05.036 · Cited by 1 · see citing works
Jørgensen, N. H., Tharmabalan, S., Aslan, I., Hansen, N. B., & Merritt, T. R. (2026) · ACM International Conference on Intelligent User Interfaces (IUI) · doi:10.1145/3742413.3789218
Unified Protocols and Pipelines: Rethinking Model Context Serving in the Era of MaaS
Ferwani, C. S., Avthankar, A., Pandey, K. K., Gongal, N., & Naik, S. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-14186-6_39
Du, H., Chen, X., Liu, Y., & Huang, K. (2026) · Wireless networks · doi:10.1007/978-3-032-20194-2_6
ContReAct: A Feedback-Based Architecture for Continuous Agentic Operation
Szeider, S. (2026) · International Workshop on Agentic Engineering · doi:10.1145/3786167.3788407
Thinking Is Not Enough: The B * Expansion Technique for Enhancing Autonomous LLM Agents
Posadas, S. A. M., & Vega, J. (2026) · Expert Systems · doi:10.1111/exsy.70333
Tamayo, J. E. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-25187-9_44
Kubam, C. S. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-25694-2_37
- Authors
- Singh, A., Ehtesham, A., Kumar, S., & Khoei, T. T.
- Affiliations
- Cleveland State University (US)
- Kent State University (US)
- Northeastern University (US)
- Venue
- Preprints.org
- Publisher
- MDPI AG
- Published
- 2025-04-02
- OpenAlex
- W4409100563
- Cited by
- 21
- References
- 0
- Citations by year
- 2026: 12 · 2025: 9
- Type
- posted content
- Language
- en
- Open access
- green
- doi.org
- Locations
- Preprints.org (acceptedVersion, cc-by)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Keywords
- Model Context Protocol, large language models, agentic AI, client-server architecture, API interoperability, security mechanisms
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
