hkqai
10 verified skills60 total stars
active-learning
Autonomous optimization using ARROWS (thermodynamically-guided synthesis) or Bayesian Optimization (generic process optimization). Orchestrates closed-loop experimentation from campaign setup through iterative learning to convergence.
documentation6
synthesis-planner
Intelligent synthesis route planning for inorganic materials. Use this skill whenever the user needs a synthesis protocol.
tools6
vasp-ase
Professional skill for setting up, executing, and debugging VASP DFT calculations using the Atomic Simulation Environment (ASE).
development6
stability-analyzer
Analyze thermodynamic stability of inorganic materials by routing between two workflows: a cheap Materials Project-backed lookup path for known compositions, and a custom self-consistent MLIP hull workflow for novel or structure-specific materials. Use this skill whenever the user asks whether a material is stable, requests energy above hull, decomposition products, polymorph context, or wants to include stability as a screening criterion. This skill is intended to become the single orchestration layer for stability analysis, with workflow branching handled in the skill rather than inside an MCP tool.
tools6
candidate-generator
**Guide materials discovery workflows from concept to crystal structures.** Generate, enumerate, and organize inorganic crystal structure candidates for screening, DFT, ML training, and experimental synthesis planning. Handles complete pipeline from elements-only input to ASE database storage with provenance tracking. **Trigger this skill for:** - Structure generation: "generate candidates", "create structures", "build crystal structures", "make supercells" - Chemical exploration: "screen compositions", "chemical substitution", "doping", "solid solutions", "ion exchange" - Materials classes: "battery cathodes", "perovskites", "isostructural analogues", "high-entropy oxides", "defects" - Configuration space: "enumerate orderings", "SQS generation", "disorder-to-ordered", "vacancy defects", "interstitials" - Scale: "generate 50 structures", "100 candidates", "high-throughput", "ML training set", "candidate library" - Starting points: Element lists ("Li-Mn-P-O system"), formulas ("LiCoO2 analogues"), structure types ("olivine"), ICSD/MP/COD IDs - Workflows: "prototype matching", "lattice perturbation", "supercell expansion" **Complete pipeline coverage:** Elements → Compositions → Seed structures → Chemical exploration → Disorder/ordering → Defects → Perturbation → ASE database **When NOT to trigger:** Reading/parsing existing structures (just use pymatgen directly), analyzing calculated properties (use candidate-screener), validating geometry (use structure_validator). **Bundled resources:** Phase guide + worked examples → references/ directory
development6
Isaac Lab Scene Initializer
Guides users through authoring or extending an Isaac Lab/Isaac Sim YAML scene file (robots, assets, positions, physics) and generates the ready-to-run `*_isaac.py` script via the bundled `isaac_scene_loader.py`. Use this skill whenever the user wants to set up a virtual robot lab scene in Isaac Sim, add or configure robots/assets in Isaac Lab, generate an Isaac Lab Python script from a YAML description, initialise a new simulation environment, or load USD models into an Isaac Sim stage — even if they don't say "YAML" or "Isaac Lab" explicitly.
development6
lula-description-generator
Generates Lula robot description YAML from a URDF file for Isaac Sim motion planning (CuMotion/RMPflow/Lula), including automatic collision sphere generation via mesh repair + voxelisation + medial-axis sphere packing
data-ai6
nsys-optimizer
Optimize CUDA/GPU simulation code using NVIDIA Nsight Systems (nsys) profiling. Use this skill whenever the user mentions performance problems, slow simulations, profiling, nsys, Nsight Systems, kernel optimization, GPU bottlenecks, or wants to speed up CUDA code. Also trigger when the user compares two scenes and one is unexpectedly slower, or asks "why is this slow?" about GPU code. This skill covers the full optimization loop: profiling, bottleneck diagnosis, targeted optimization, verification, and iterative measurement.
development6
URDF Validator & Fixer
Pre-import validation and auto-fix for URDF files targeting Isaac Sim / USD
tools6
candidate-screener
Transform candidate structures into ranked, property-enriched, synthesis-ready materials through intelligent validation, hierarchical property retrieval, and multi-objective optimization. Validates structure integrity, retrieves properties via intelligent hierarchy (Materials Project DFT → ASE cache → ML prediction with MatGL/matcalc), handles disordered structures through preprocessing layer (majority/enumeration/SQS ordering), applies application-specific screening criteria, and ranks by confidence-weighted multi-objective methods. Core strength: Complete transparency - tracks rejection reasons, flags ML uncertainties for DFT verification, implements confidence scoring, and provides complete provenance. Supports all screening workflows: battery cathodes (voltage, capacity, stability), catalysts (surface energies, adsorption), thermoelectrics (band gap, transport), phosphors (optical properties, rare-earth incorporation), mechanical materials (elasticity, phonons), and custom multi-criteria screening. When NOT to trigger: simple property lookup without ranking/filtering (use MP/ASE tools directly), VASP/DFT result analysis (use vasp-ase skill), active learning with Bayesian optimization (use active-learning skill). Trigger keywords: screen candidates, validate structures, property retrieval, hierarchical data, rank materials, filter candidates, multi-objective optimization, battery screening, catalyst discovery, thermoelectric materials, phosphor screening, mechanical properties, phonon stability, surface energies, formation energy prediction, band gap prediction, disordered structures, dilute doping, solid solutions, materials discovery pipeline, high-throughput screening, candidate ranking, confidence scoring, MP + ML hybrid, MatGL predictions, matcalc calculations, structure preprocessing, ordering strategies, synthesis-ready candidates.
tools6