Claude Skills by VectorSpaceLab

"Work with MONAI Bundle configuration syntax, metadata/spec files,

"Build MONAI preprocessing, IO, metadata, caching, dataset,

"Choose and wire MONAI networks, losses, metrics, inferers,

"Assemble MONAI training and evaluation loops with Ignite-based

"Use MonoGS for CUDA Gaussian-splatting SLAM, dataset/config setup,

"Acquire MonoGS datasets, validate dataset layouts, and edit

Install MonoGS, build its CUDA extensions, and verify backend readiness.

"Evaluate MonoGS runs, inspect saved result trees, and handle

Run and troubleshoot MonoGS RealSense live demos and the Open3D/OpenGL GUI.

"Run MonoGS offline SLAM on monocular TUM, RGB-D TUM/Replica, and

"Route MOSS-TTS family speech, voice-agent, sound-effect,

"Prepare and validate MOSS-TTS fine-tuning JSONL data and launch

"Operate Hugging Face remote-code and Gradio-style workflows for

Operate the torch-free and low-memory MOSS-TTS-Delay llama.cpp backend.

"Operate MOSS-TTS Local Transformer v1.5 batch and realtime

"Operate MOSS-TTS-Realtime low-latency voice-agent streaming,

"Operate MOSS-SoundEffect v2 DiT/DAC/Qwen3 inference, demo,

"Routes OpenMOSS/MOSS workflows for local LLM inference, model

"Routes MOSS SFT data and fine-tuning tasks for conversation

"Routes MOSS chat inference tasks for prompt formatting, PyTorch

"Routes MOSS model-runtime tasks for config, tokenizer, causal

"Routes MOSS serving and UI deployment tasks for FastAPI payloads,

"Use Motus, a unified latent-action world model, for

"Prepare, validate, convert, and batch Motus robot, latent-action,

"Run or plan Motus CUDA inference for real-world images and

"Configure, validate, launch, and resume Motus three-stage training

"Use ms-swift for LLM and multimodal training, inference,

"Plan and debug advanced ms-swift RLHF, GRPO/GKD, rollout, Ray, and

"Customize ms-swift datasets, registries, plugins, models, and

Export, quantize, push, and EvalScope-evaluate ms-swift models safely.

"Run and debug ms-swift inference, app, deployment, backend

"Build, debug, and explain ms-swift pre-training and supervised

"Use MTEB to evaluate embedding models, select tasks and

"Use the MTEB console script for safe CLI workflows, shell

"Add or review MTEB task, benchmark, model metadata,

"Run MTEB evaluations from Python, configure

"Load and validate MTEB models, built-in model metadata,

"Load, inspect, validate, submit, and display MTEB result caches,

"Discover, filter, inspect, and reason about MTEB tasks and

"Use MuJoCo Menagerie robot MJCF assets: choose models, load XML

"Contributor and maintainer workflow guidance for MuJoCo Menagerie

"Choose MuJoCo Menagerie model directories, scene XMLs, variants,

"Guide safe Menagerie MJCF editing, attachment composition,

"Load, inspect, validate, and lightly simulate Menagerie MJCF XMLs

"Use the Tiled Diffusion & VAE AUTOMATIC1111 WebUI extension for

"Configure and troubleshoot the extension's DemoFusion panel for

"Configure and troubleshoot the extension's Tiled Diffusion panel

"Configure and troubleshoot the extension's Tiled VAE panel for

"Route NVlabs MUNIT legacy multimodal image-to-image translation

"Validate MUNIT dataset layouts, YAML configs, demo-data