AI4S-agent-tools
Collecting a variety of Agent-Ready tool modules。主要特性包括：Materials Science、DPACalculator - Deep learning atomistic simulations with universal potentials、Thermoelectric - Materials screening with CALYPSO。还包含 2 个其他特性。
详细介绍
项目简介
Collecting a variety of Agent-Ready tool modules。主要特性包括：Materials Science、DPACalculator - Deep learning atomistic simulations with universal potentials、Thermoelectric - Materials screening with CALYPSO。还包含 2 个其他特性。
主要特性
- •Materials Science
- •DPACalculator - Deep learning atomistic simulations with universal potentials
- •Thermoelectric - Materials screening with CALYPSO
- •ABACUS-tools - First-principles calculations with ABACUS
- •Chemistry
- •PubChem - Compound data retrieval and structure download
适用场景
快速开始
# 克隆仓库
git clone https://github.com/deepmodeling/AI4S-agent-tools
# 进入目录
cd AI4S-agent-tools
# 查看文档
cat README.md原文 README
AI4S-agent-tools
An open project by the DeepModeling community - Building intelligent tools for scientific research.
🌐 View Tool Showcase |
🤝 Contribute |
🎯 Mission
We're building a comprehensive "scientific capability library" - agent-ready tools that cover the full spectrum of AI for Science tasks:
🔬 Current Tools
Materials Science
- DPACalculator - Deep learning atomistic simulations with universal potentials
- Thermoelectric - Materials screening with CALYPSO
- ABACUS-tools - First-principles calculations with ABACUS
Chemistry
Research Tools
- Paper Search - ArXiv literature search and parsing
- DeepMD Docs RAG - Documentation knowledge base
🚀 Coming Soon
- 📊 Spectral analysis (XRD, NMR, Raman)
- 🧬 Protein structure prediction
- 🔭 3D molecular visualization
- 📈 Experimental design optimization
- 🧫 Multi-objective Bayesian optimization
💻 Quick Start
Use a Tool
# Install dependencies
cd servers/<tool-name>
uv sync
# Run the server
python server.py --port 50001
Add Your Tool
# Copy template
cp -r servers/_example servers/my_tool
# Edit and test
cd servers/my_tool
# ... edit server.py ...
uv sync
python server.py --port 50002
🆕 AI-Powered Tool Generator (Beta)
Generate MCP tools automatically through conversational AI:
# Configure environment variable (create .env file)
# Run the interactive tool generator
adk web
# Follow the conversational prompts to create your tool
The MCP Agent will guide you through creating a fully-configured MCP server that follows all project standards. This beta feature uses AI to:
- Collect your tool requirements through friendly dialogue
- Generate complete server code with proper error handling
- Create all necessary configuration files (metadata.json, pyproject.toml)
- Ensure compliance with project standards
See mcp-agent/README.md for more details.
See CONTRIBUTING.md for the complete guide.
🏗️ Architecture
Each tool is an independent MCP (Model Context Protocol) server that can be:
- Run standalone for development
- Integrated with AI agents (Claude, GPT, etc.)
- Composed into complex workflows
🤝 Join Us
We welcome contributions from:
- 🧑🔬 Domain scientists with computational needs
- 💻 Developers interested in scientific computing
- 🤖 AI researchers building science agents
- 📚 Anyone passionate about open science
💬 Community
Join our WeChat community group to discuss ideas and collaborate:
📄 License
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.
🙏 Acknowledgments
Built with ❤️ by the DeepModeling community.