About Flow
Flow is an MCP server published by wwh0411 in the Developer Tools category: [ACL 2026 Main] MCP-Flow: Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP Tools.
The repository has 26 stars and 3 forks, with the last commit 6 months ago. Six months or more without a commit doesn't mean the server is broken, but check the open issues before depending on it in production.
Install
claude mcp add flow -- npx -y @outblock/flow-mcpnpx -y @outblock/flow-mcpThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Ask AI
Ask AI about Flow
Powered by Claude · Grounded in docs
Security checks
Static checks derived from the repository 4 days ago. Heuristics on the source, not an audit.
- ✓README presentThe repository ships documentation.
- !License declaredNo license file. Legally you have no right to use it in a product.
- !Tests presentNo tests found. Behaviour changes may go unnoticed between versions.
- !Dependencies pinnedNo lockfile. Installs can pull different code over time.
- !No dynamic code executionSource scan found eval or shell execution. Review how tool inputs reach it before granting credentials.
- !Scoped permissionsDoesn't declare a permission scope. Assume it can do anything its process can.
README
MCP-Flow
Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP Tools
🗓️ News
- 🎉 Apr 6, 2026 — We are happy to announce that MCP-Flow has been accepted to the Main Conference of ACL 2026!
- 🧠 Oct 28, 2025 — MCP-Flow is released on arXiv.
- 🛠️ Nov 10, 2025 — We open-source all the server configurations and tool information!
- 🛠️ Nov 28, 2025 — We open-source all the instruction-function call pairs and the testsets for evaluation inluding in-domain and OOD settings! The dataset is available at HuggingFace
📝 Introduction
MCP-Flow is an automated web-agent-driven pipeline for large-scale server discovery, data synthesis, and model training in the Model Context Protocol (MCP) ecosystem.
🌐 Key Features
-
🤖 Automated server collection from 6 major MCP marketplaces
-
📊 Extensive tool coverage: 1,166 real-world servers, 11,536 tools, and 68K+ instruction–function call pairs
-
🧩 Scale & diversity far beyond previous benchmarks
📂 Datasets
| Category | Path | Description |
|---|---|---|
| 🧠 Function calls & trajectories | ./data/function_call/ & ./data/trajectory/ |
Example data; full datasets are released on HuggingFace |
| ⚙️ MCP configurations | ./data/mcp_config/ |
Configuration files for discovered servers |
| 🧰 Tool information | ./data/tools/ |
Tool descriptions and schema definitions |
| 💻 Source code | ./src/ |
Core scripts for server deployment |
| 📲 Testset | ./test_data on HuggingFace |
Including in-domain and out-of-domain evaluation settings |
🛠️ Installation
git clone https://github.com/<your-org>/MCP-Flow.git
cd MCP-Flow
pip install -r requirements.txt
🧾 Citation
If you find MCP-Flow useful in your research, please consider citing:
@misc{wang2025mcpflowfacilitatingllmagents,
title={MCP-Flow: Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP Tools},
author={Wenhao Wang and Peizhi Niu and Zhao Xu and Zhaoyu Chen and Jian Du and Yaxin Du and Xianghe Pang and Keduan Huang and Yanfeng Wang and Qiang Yan and Siheng Chen},
year={2025},
eprint={2510.24284},
archivePrefix={arXiv},
primaryClass={cs.AI},
url={https://arxiv.org/abs/2510.24284},
}
📧 Contact
If you have any questions or encounter issues, feel free to open an issue or reach out to the authors directly:
📮 Email: 12321254@zju.edu.cn
💬 WeChat:
README fetched from the repository 4 days ago. The original is authoritative.