About Atomistic Toolkit MCP Server
An MCP server for atomistic modeling workflows, built around ASE, pymatgen, and modern machine-learning interatomic potentials. It exposes structure preparation, optimization, molecular dynamics, and analysis as MCP tools, with downloadable file artifacts.
Key features
- 01Builds or imports atomistic structures
- 02Edits cells and creates defects
- 03Runs geometry optimization and molecular dynamics
- 04Estimates bulk modulus and analyzes structures or trajectories
- 05Produces downloadable data files and plots
Use cases
- 01Create bulk, surface, or molecule structures
- 02Add vacancies or substitutions to a structure
- 03Run optimization and NVE, NVT, or NPT simulations
- 04Analyze RDF, MSD, VACF, and diffusion
MCP support
FastMCP exposes structure operations, optimization, simulation, and analysis as tools for MCP clients. It supports Streamable HTTP and STDIO, and includes an artifact download endpoint.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEMCP Atomic Toolkit
Note
This project is under active development. Interfaces and behavior may evolve.
A FastMCP server for atomistic modeling workflows powered by ASE, pymatgen, and modern ML interatomic potentials.
It gives MCP clients a practical toolkit for:
- building and importing structures,
- editing cells (strain, supercell) and defects (vacancy, substitution, interstitial),
- running geometry optimization + NVE/NVT/NPT molecular dynamics,
- estimating bulk modulus,
- analyzing structures/trajectories,
- and downloading generated artifacts (data + plots).
✨ Why this repo
If you need atomistic workflows exposed as MCP tools (instead of hand-wiring scripts), this project gives you:
- ready-to-call MCP tools for common simulation tasks,
- file-first outputs that are easy to inspect/reuse,
- artifact download URLs so clients don’t need binary blobs in chat context,
- deployment-ready HTTP app with health and server-card endpoints.
🚀 Features
- MCP-native workflows via FastMCP tools
- Structure generation: bulk, surface, molecule, supercell, amorphous, liquid, bicrystal, polycrystal
- Structure import from xyz/cif/POSCAR text
- Edits and defects: rotate, translate, strain, supercell, wrap, vacancy, substitute, interstitial
- Calculators:
autotrieskim→orb→nequix→ ASE EMT (Al/Cu/Ag/Au/Ni/Pd/Pt) - Analytical fits: Lennard-Jones, Morse, and Buckingham pair potentials (Render-friendly)
- Optimization and single-point energy/forces/stress
- Molecular dynamics: Velocity Verlet, Langevin, NVT Berendsen, NPT Berendsen
- Recipe tool
relax_and_md_workflow(relax then MD) - Isotropic bulk modulus from a 5-point energy-vs-strain fit
- Analysis outputs: RDF + coordination, MSD + thermo trends, VACF + diffusion
- Downloadable artifacts (
xyz,extxyz,cif,traj,png,svg,csv,dat, ...) - Registry-friendly endpoints (
/healthz, server card, Streamable HTTP root)
⚡ Quick Start
1) Requirements
- Python 3.11+
2) Install
Core install does not require the OpenKIM C++ API:
pip install -r requirements.txt
or:
pip install -e .
OpenKIM (kimpy) is optional. It compiles against the system KIM API, so a default install used to fail on machines without libkim-api.
To enable the KIM calculator:
# Debian/Ubuntu
sudo apt-get install -y libkim-api-dev pkg-config
pip install -e ".[kim]"
macOS (Homebrew): brew install openkim-models kim-api then pip install -e ".[kim]".
Without the extra, use calculator_name='auto', 'emt', 'orb', or 'nequix'. Runtime code already falls back when KIM is missing. EMT covers Al, Cu, Ag, Au, Ni, Pd, and Pt.
3) Run locally
uvicorn mcp_atomictoolkit.http_app:app --host 0.0.0.0 --port 10000
Alternative:
python main.py
STDIO mode (for desktop MCP clients):
python -m mcp_atomictoolkit.mcp_server
Important
STDIO transports must keep stdout clean for JSON-RPC. Avoid
print()or logging to stdout when running the server in STDIO mode.
4) Smoke check
curl -s http://localhost:10000/healthz
Expected response:
{"status":"ok"}
🧰 Tooling Overview
Main MCP tools exposed by the server:
list_workspace_capabilities_workflowbuild_structure_workflowimport_structure_workflowmanipulate_structure_workflowanalyze_structure_workflowwrite_structure_workflowoptimize_structure_workflowsingle_point_workflowestimate_elastic_workflowrun_md_workflowrelax_and_md_workflowanalyze_trajectory_workflowautocorrelation_workflow
Legacy aliases are also registered: build_structure, read_structure_file, write_structure_file, optimize_with_mlip.
Call list_workspace_capabilities_workflow first from an agent. It reports which calculators imported, EMT element support, integrators, structure types, and edit operations.
🌐 Endpoints
POST /— primary MCP Streamable HTTP endpointGET /healthz— health checkGET /docs— lightweight documentation (README)GET /.well-known/mcp/server-card.json— MCP server card metadataGET /artifacts/{artifact_id}/{filename}— artifact download route/sse/— compatibility alias path mounted to the MCP app
📦 Deployment
Render
render.yaml is included and ready to use.
Default start command:
uvicorn mcp_atomictoolkit.http_app:app --host 0.0.0.0 --port $PORT
Docker
The image installs libkim-api-dev and the optional [kim] extra.
docker build -t mcp-atomictoolkit .
docker run --rm -p 7860:7860 mcp-atomictoolkit
🗂️ Project Structure
src/mcp_atomictoolkit/
mcp_server.py # FastMCP tool definitions
http_app.py # Starlette app + routing/endpoints
workflows/core.py # High-level workflow orchestration
analysis/ # Structure/trajectory/VACF analysis logic
structure_operations.py
optimizers.py
md_runner.py
artifact_store.py # Download artifact registration + URLs
🧪 Workflow Notes (for MCP clients)
Structure building coverage
build_structure_workflow supports:
- bulk (ASE
bulk) - surface (ASE
surface) - molecule (ASE
molecule) - supercell (multiplication of a base structure)
- amorphous/liquid (random packed structures)
- bicrystal and polycrystal (grain stacking/rotation)
Paste an existing geometry with import_structure_workflow (xyz, cif, or poscar text).
Edits and defects
manipulate_structure_workflow operations:
rotate,translate,strain,supercell,wrapvacancy(operation_kwargs.index)substitute(index,symbol)interstitial(symbol, optionalposition)
Optimization options
optimize_structure_workflow exposes:
max_steps,fmax(convergence)maxstep,alpha(BFGS step/damping controls)constraints(fixed_atoms,fixed_bonds,fixed_cell)
Single-point and elasticity
single_point_workflow computes energy, forces, and stress (if periodic).
estimate_elastic_workflow fits E(strain) at five isotropic strains and returns bulk_modulus_GPa plus calculator_used.
MD integrators / ensembles
run_md_workflow supports:
velocityverlet/nve(NVE)langevin/nvt-langevin(NVT)nvt/nvt-berendsen(NVT)npt/npt-berendsen(NPT;pressure_GPa,taup)
relax_and_md_workflow chains optimization then MD.
Every energy/MD result includes calculator_requested, calculator_used, and calculator_fallbacks so an EMT copper run is not mistaken for an MLIP result.
📈 GitHub Pulse
Star history
🤝 Contributing
- Keep outputs file-based and artifact-friendly.
- When adding tools, usually update both:
workflows/core.pymcp_server.pyhttp_app.pyTOOL_NAMES
- Preserve
http_app.pycompatibility behavior unless intentionally changing deployment contracts.
📄 License
MIT — see LICENSE.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.