hpc-mcp
Provides MCP tools for high-performance computing (HPC) and converged computing, enabling agentic interactions with HPC resources and workflows.
README
hpc-mcp
🌀 Agentic MCP tools for High Performance Computing

Related Projects
- flux-mcp: MCP functions for Flux Framework.
- fractale: (fractale) MCP orchestration (agents, databases, ui interfaces).
Usage
This is a library of MCP tools (functions, prompts, and resources) intended for converged computing and HPC use cases. A demo server is provided here, and largely functions are expected to be used a-la-carte as imports to other libraries. We welcome contributions of all functions types that are related to HPC, converged computing, and science. These MCP tools can be used via a standalone server, or combined with other tools.
Environment
Control of tools (access, limits, etc.) is done by way of the environment.
| Name | Description | Default |
|---|---|---|
HPCMCP_FILESYSTEM_SANDBOX |
enable (require) a sandbox | False |
HPCMCP_FILESYSTEM_DATA_ROOT |
if sandbox enabled, read-only data root | unset |
HPCMCP_FILESYSTEM_RESULT_ROOT |
if sandbox enabled, read and write result root for agent to write | unset |
HPCMCP_FILESYSTEM_TOKEN_LIMIT |
Hard limit of tokens to allow in response | unset |
Server
We provide examples for fastmcp and a vanilla mcp (stdio) setup. Neither requirements are added to the install directly, so it's up to the user (you) to install. Tests are performed with fastmcp (TBA)
fastmcp
You will need fastapi and fastmcp installed.
# fastmcp
pip install fastmcp fastapi
To start the demo server:
# Vanilla MCP (with cli)
echo '{"jsonrpc": "2.0", "id": 1, "method": "initialize", "params": {"protocolVersion": "2024-11-05", "capabilities": {}, "clientInfo": {"name": "test", "version": "1.0"}}}' | python3 -m hpc_mcp.server | jq
# Initialize and list tools
(echo '{"jsonrpc": "2.0", "id": 1, "method": "initialize", "params": {"protocolVersion": "2024-11-05", "capabilities": {}, "clientInfo": {"name": "manual-test", "version": "1.0"}}}';
echo '{"jsonrpc": "2.0", "method": "notifications/initialized"}';
echo '{"jsonrpc": "2.0", "id": 2, "method": "tools/list"}') | python3 -m hpc_mcp.server | jq
# FastMCP
python3 -m hpc_mcp.server.fastmcp
Testing
I will add tools to git as I write tests for them. To test, start the fastmcp server in one terminal:
python3 -m hpc_mcp.server.fastmcp
In another terminal, run the test. You'll need to pip install pytest pytest-asyncio
pytest -xs tests/test_build_docker.py
# or
pytest -xs tests/test_*.py
TODO
- Add annotated descriptions to all functions for LLM.
Tools to add:
- helpers
- debug
- result parser (regular expressions)
- kubernetes
- deploy job
- deploy minicluster
License
DevTools is distributed under the terms of the MIT license. All new contributions must be made under this license.
See LICENSE, COPYRIGHT, and NOTICE for details.
SPDX-License-Identifier: (MIT)
LLNL-CODE- 842614
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。