Genkai_Agent

Prepare, validate, and submit UMA/fairchem machine-learning interatomic potential calculations on Genkai with the established PJM launcher and UMA virtual environment. Use for MLIP or UMA structure relaxation, molecular dynamics, energy/force inference, GPU calculations, PJM submission, restart preparation, or locating and reporting calculation outputs.

Read surface-research PDFs or JSON records and extract structured reaction, material, and modeling information for downstream agent workflows.

Build realistic oxide-surface candidates: oxygen-vacancy landscapes, adsorbate coverage landscapes, and metal-cluster-on-surface starting structures using ASE, Optuna, and optional UMA/FAIRChem calculators.

Filter and normalize paperread surface-extraction outputs into Agent-ready modeling inputs, including facet equivalence, loaded nanoparticle species, material class, reaction type, and executable modeling task mapping.

Prepare, stage, and summarize LOBSTER bonding-analysis calculations from completed VASP, CP2K, or Quantum ESPRESSO outputs.

Skill for executing ABACUS DFT materials calculations. Help users set up, run, and analyze ABACUS calculations.

Skills for running ASE molecular dynamics (MD) and structure optimisation using a DeePMD interatomic potential — input preparation, remote batch submission to Bohrium, and result collection.

Build, inspect, modify, and curate atomic structures using inline Python code with ASE (no external scripts required).

Submit and manage computational jobs on Bohrium (bohrium.com) cloud platform using bohr CLI. Recommended for submission methods on Bohrium.

Predict material properties of crystal structures with CGCNN models.

Concept skill for Density Functional Theory (DFT) calculations. Describes what DFT is, when to use it, and which tool skills to invoke for DFT jobs. Use this to understand the DFT landscape before selecting a specific code (VASP or ABACUS).

Concept skill for Machine Learning Force Fields (MLFFs). Describes what MLFFs are, the distinction between training and inference, and which tool skills to use. Load this before selecting a specific MLFF framework (DeePMD, MatterSim, etc.).

Concept skill for Molecular Dynamics (MD) simulation. Describes MD ensembles, key parameters, and which tool skills to invoke for MD runs. Use this to understand MD concepts before selecting a specific simulation tool.

Concept skill for crystal structure generation and atomic structure manipulation. Covers generative models for inverse design as well as manual structure building. Use this to understand the structure generation landscape before selecting a specific tool.

Concept skill covering general-purpose utility capabilities: visualization/plotting, structure format conversion, database queries, and web search. Use this to identify which utility tool skill to load for a given support task.

Skills for materials dataset operations using CLI tools. Query and manage ASE databases and a normalized SQLite information database (nodes/datasets/dataset_elements schema). Includes SQL composition guidance for query-info.

DeePMD-kit training, finetuning, testing, and model inspection skill. Use this skill whenever training or finetuning a Deep Potential (DP / DPA-1 / DPA-2) model, running model tests, or inspecting model parameters. Training is split into a preparation phase (data conversion + input.json generation, always local) and an execution phase (dp CLI commands, local or via dpdisp skill on hpc or Bohrium).

Run Shell commands as computational jobs, on local machines or HPC clusters, through Shell, Slurm, PBS, LSF, Bohrium, etc. USE WHEN the user needs to submit batch jobs to a cluster, run commands on a remote server, execute tasks via job schedulers (Slurm, PBS, LSF), or safely run long-term/background shell commands that require state tracking and auto-recovery.

General guideline for MLFF model training. Describes the decision flow for choosing between pre-trained model validation, fine-tuning on available data, and launching a full PFD workflow.

Iterative workflow for materials design through candidate generation, property prediction, screening, and refinement.

Iterative workflow for model distillation. Generates a light ML force field from a pre-trained or fine-tuned model via active learning.

Iterative workflow that fine-tunes pre-trained ML force fields using DFT-labeled data via active learning.

LAMMPS molecular dynamics with DeepMD/DPA machine-learning potentials ONLY. This skill generates LAMMPS input files (in.lammps) using pair_style deepmd. Classical force fields (Stillinger-Weber, Tersoff, EAM, etc.) are NOT supported and must NEVER be suggested. The default model is the DPA3 pretrained checkpoint frozen with the Omat24 head (from DEEPMD_MODEL_PATH env var), identical to the ase-deepmd skill. A model file is always required — do NOT ask the user whether they want to use a classical potential instead. Bohrium's default DeepMD image (registry.dp.tech/dptech/deepmd-kit) already includes LAMMPS — no separate LAMMPS installation is needed. Covers structure conversion, input generation, Bohrium submission via bohrium skill, and result collection.

A modular Python framework for building and analyzing atomic structures with CLI tools for materials modelling.

Query Materials Project data for crystal structures, stability, electronic properties, phase diagrams.

Evaluate MatterGen-generated crystal structures for novelty, uniqueness, stability, and S.U.N. criteria, and produce filtered relaxed structures and evaluation reports.

Finetune MatterGen models and produce a reusable finetuned checkpoint for downstream crystal generation.

Generate inorganic crystal structures with MatterGen, supporting unconditional sampling, property-conditioned generation, and sampling from custom trained or finetuned checkpoints.

MatterGen is a diffusion-based generative model for inorganic crystal structure generation. Use this skill to understand what MatterGen can do and which sub-skill to load. For generation, fine-tuning, or evaluation tasks, load the appropriate sub-skill.

Run MatterSim workflows for batch structure relaxation, molecular dynamics, model finetuning, and post-MD ionic conductivity analysis on crystal structure datasets..

Skill for scientific plotting using matplotlib. Generates, executes, and verifies publication-quality plots directly via run_python and run_bash.

Calculate entropy of atomic structure descriptors and select maximally diverse subsets for active learning using the QUEST method.

Convert crystal structure files between extxyz and CIF, including splitting multi-frame extxyz into per-frame CIF files.

Tavily is an AI-powered web intelligence platform providing search, content extraction, site crawling, URL mapping, and deep research. Use this skill to understand what Tavily can do and which sub-skill to load for a given web task.

Web search, content extraction, crawling, and deep research via the Tavily CLI. Use this skill whenever the user wants to search the web, find articles, research a topic, look something up online, extract content from a URL, grab text from a webpage, crawl documentation, download a site's pages, discover URLs on a domain, or conduct in-depth research with citations. Also use when they say "fetch this page", "pull the content from", "get the page at https://", "find me articles about", or reference extracting data from external websites. This provides LLM-optimized web search, content extraction, site crawling, URL discovery, and AI-powered deep research — capabilities beyond what agents can do natively. Do NOT trigger for local file operations, git commands, deployments, or code editing tasks.

Crawl websites and extract content from multiple pages via the Tavily CLI. Use this skill when the user wants to crawl a site, download documentation, extract an entire docs section, bulk-extract pages, save a site as local markdown files, or says "crawl", "get all the pages", "download the docs", "extract everything under /docs", "bulk extract", or needs content from many pages on the same domain. Supports depth/breadth control, path filtering, semantic instructions, and saving each page as a local markdown file.

Programmatic web search with context isolation. Use this skill for any research task where you need to search the web, filter results, and extract specific information — without polluting your context window with raw HTML and boilerplate. This is the default skill for web research. Triggered by "search for", "look up", "find", "research", "what's the latest on", or any query that requires current web information. Also use when asked to "search and filter", "find the important parts", or "extract the key details" — any case where the user wants curated, noise-free content.

Extract clean markdown or text content from specific URLs via the Tavily CLI. Use this skill when the user has one or more URLs and wants their content, says "extract", "grab the content from", "pull the text from", "get the page at", "read this webpage", or needs clean text from web pages. Handles JavaScript-rendered pages, returns LLM-optimized markdown, and supports query-focused chunking for targeted extraction. Can process up to 20 URLs in a single call.

Discover and list all URLs on a website without extracting content, via the Tavily CLI. Use this skill when the user wants to find a specific page on a large site, list all URLs, see the site structure, find where something is on a domain, or says "map the site", "find the URL for", "what pages are on", "list all pages", or "site structure". Faster than crawling — returns URLs only. Essential when you know the site but not the exact page. Combine with extract for targeted content retrieval.

Conduct comprehensive AI-powered research with citations via the Tavily CLI. Use this skill when the user wants deep research, a detailed report, a comparison, market analysis, literature review, or says "research", "investigate", "analyze in depth", "compare X vs Y", "what does the market look like for", or needs multi-source synthesis with explicit citations. Returns a structured report grounded in web sources. Takes 30-120 seconds. For quick fact-finding, use tavily-search instead.