The Cost of Saying Hello: An Empirical Study of Handshake Overhead in AI Agent Communication Protocols
Source
arxiv.org
Author
Pranav Varshney
Date
Why it matters
Quantifies protocol overhead: a legacy MCP handshake takes three round trips and 230 ms at 50 ms RTT versus 61 ms for A2A, and LLM-read metadata adds token cost.
Key takeaways · AI-distilled
The benchmark instruments each protocol's official SDK, covering MCPThe Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.Full definition →'s 2025 initialize and 2026 discover eras, A2A and ACP, across cold-start, warm-repeat, capability-mismatch and degraded-metadata cases, 50 runs per cell on loopback and at an emulated 50 ms RTT.
When an LLMA large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.Full definition → reads the counterpart's metadata, handshake cost scales with its capability inventory at 24-107 tokens per capability, so a 200-capability counterpart uses 11% of a 200k-token context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition → before the AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → reads its own task.
MCP's 2026 revision cuts its handshake from three round trips to two, and with era pinning to zero on warm reconnects, which the author reads as protocol evolution corroborating the measured cost.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
articleWhen Agent Metrics Measure Different Things: An Evidence-Grounded Audit of the Praxa AI PipelineStefan G. Creadore, Peyton Woakz
articleIs Bash All You Need? An Empirical Study of Tool Interfaces for Enterprise Digital Worker AgentsHazel Mak, Susheel Suresh, Sahil Bhatnagar, Barry Wang, Chhaya Methani, Alejandro Gutierrez Munoz
articleA Few Pages of Markdown: Committed AI Configuration and Lower Quality Cost after Coding-Agent AdoptionYegor Denisov-Blanch, Shyam Agarwal, Pavel Azaletskiy, Hao He, Rylan Schaeffer, Brando Miranda, Bogdan Vasilescu, Sanmi Koyejo