MCP-Flow: Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP Tools

TL;DR

MCP-Flow is an automated pipeline that enables large language models to better utilize diverse MCP tools by discovering servers, synthesizing data, and training models at scale.

Contribution

It introduces MCP-Flow, a scalable, automated system for large-scale MCP data collection, training, and tool utilization, surpassing prior efforts in scale and diversity.

Findings

MCP-Flow collected data from over 1,166 servers and 11,536 tools.

Produced 68,733 high-quality instruction-function call pairs.

Demonstrated improved MCP tool selection and task performance.

Abstract

Large Language Models (LLMs) increasingly rely on external tools to perform complex, realistic tasks, yet their ability to utilize the rapidly expanding Model Contextual Protocol (MCP) ecosystem remains limited. Existing MCP research covers few servers, depends on costly manual curation, and lacks training support, hindering progress toward real-world deployment. To overcome these limitations, we introduce MCP-Flow, an automated web-agent-driven pipeline for large-scale server discovery, data synthesis, and model training. MCP-Flow collects and filters data from 1166 servers and 11536 tools, producing 68733 high-quality instruction-function call pairs and 6439 trajectories, far exceeding prior work in scale and diversity. Extensive experiments demonstrate MCP-Flow's effectiveness in driving superior MCP tool selection, function-call generation, and enhanced agentic task performance. MCP-Flow thus provides a scalable foundation for advancing LLM agents' proficiency in real-world MCP environments. MCP-Flow is publicly available at https://github.com/wwh0411/MCP-Flow.