MLOps
模型部署、评估与运维
slurm-gpu-training
dongzhuoyao
Use when running ML training on HPC clusters with Slurm, including job submission, environment setup, monitoring, and failure triage. Applies to any GPU training workload on Slurm-managed clusters. Triggers: "sbatch", "srun", "Slurm", "SBATCH", "job submission", "HPC", "cluster", "walltime", "squeue"
multi-model-meta-analysis
petekp
Synthesize outputs from multiple AI models into a comprehensive, verified assessment. Use when: (1) User pastes feedback/analysis from multiple LLMs (Claude, GPT, Gemini, etc.) about code or a project, (2) User wants to consolidate model outputs into a single reliable document, (3) User needs conflicting model claims resolved against actual source code. This skill verifies model claims against the codebase, resolves contradictions with evidence, and produces a more reliable assessment than any single model.
gpu-training-acceleration
dongzhuoyao
Use when optimizing PyTorch training speed or memory on CUDA GPUs — global flags, torch.compile, fused optimizers, mixed precision, gradient checkpointing, kernel fusion, memory layout, or latent-space training. Applies to any PyTorch training workload. Triggers: "torch.compile", "TF32", "fused optimizer", "mixed precision", "bf16", "fp16", "gradient checkpointing", "Triton kernel", "CUDA flags", "GPU slow", "GPU memory"
fail-fast-ml-engineering
dongzhuoyao
Use when designing ML training pipelines, data loaders, or inference systems. Enforces engineering discipline — no silent fallbacks, explicit errors on critical paths, config as single source of truth. Triggers: "silent failure", "fallback", "preflight", "assertion", "error handling", "fail fast", "config truth"
model-first-reasoning
petekp
Apply Model-First Reasoning (MFR) to code generation tasks. Use when the user requests "model-first", "MFR", "formal modeling before coding", "model then implement", or when tasks involve complex logic, state machines, constraint systems, or any implementation requiring formal correctness guarantees. Enforces strict separation between modeling and implementation phases.
wandb-experiment-tracking
dongzhuoyao
Use when integrating W&B experiment tracking into ML training pipelines, including logging strategy, run configuration, and online/offline mode management. Triggers: "W&B", "wandb", "weights and biases", "experiment logging", "wandb.log", "wandb.init", "training dashboard"
generate-worldgen
jujumelona
Generate only explicitly requested mod-owned structures, biomes, dimensions and feature resources inside the
runtime-playtest
jujumelona
Run a disposable server/client for the approved artifact and collect bounded, observable runtime evidence.
eval
williamwue
"Compare workflow variants with blinded outputs and a frozen rubric."
eval
williamwue
"Run a blinded, repeatable comparison of workflow or model variants using real artifacts and a frozen rubric."
ai-engineer
aibangjuxin
You are a highly skilled AI Engineer specializing in the practical application of machine learning models. You are an expert in Python and popular AI/ML frameworks like TensorFlow, PyTorch, and scikit-learn. You excel at data preprocessing, model training, evaluation, and deployment.
statistical-analysis
YuukiAS
Plan and report statistical analyses: estimand framing, test/model choice, assumption checks, power/sensitivity notes, effect sizes, uncertainty, and reproducible reporting.
release-security
jujumelona
Release only artifact-identical validated source, JAR, resources, provenance, licenses, and receipts.
generate-fabric-core
jujumelona
Generate the approved Fabric target core project and registrations.
select-compatible-ai-technique
jujumelona
Research, compare, and fail-closed gate request-derived AI, agent, speech, translation, or consented voice-adaptation
principle-never-block-on-the-human
williamwue
"Apply when tempted to ask 'should I do X?' on reversible work. Proceed, present the result, let the human course-correct after the fact; reserve confirmation for irreversible actions."
principle-model-the-domain
williamwue
"Apply when writing stateful logic, or when code branches a lot or repeats a shape assumption across files. Encode the domain in a structure instead of scattered conditionals."
inspect-existing-project
jujumelona
Inventory an existing source/release archive and index its source without execution.
bayesian-modeling
YuukiAS
Bayesian and statistical modeling workflows with PyMC, diagnostics, simulation, and model reporting.
tech-research-pipeline
tomwangowa
Orchestrates a full technical research workflow by chaining specialized skills in sequence: brainstorming → tech-feasibility → assumption-extractor → micro-poc-validator → critical-research → narrative-auditor → research-cross-validator → research-synthesis. Use when evaluating a technology choice, planning a migration, or making any significant technical decision that warrants rigorous multi-angle analysis. Triggered by "full research pipeline", "rigorous tech evaluation", "research pipeline", or "evaluate [tech] thoroughly".
XState v5 Expert Skill
boneskull
Expert guidance for implementing and analyzing XState v5 state machines with TypeScript support and best practices
statsmodels
YuukiAS
Implement and review statsmodels analyses: OLS/GLM/mixed models, time series, robust covariance, diagnostics, coefficient tables, and reproducible inference workflows.
content-modeling-best-practices
display-design-studio
Structured content modeling principles with Sanity-specific guidance. Use when designing or reviewing content schemas, document types, field structures, or content architecture — including decisions around references vs embedding, content reuse, taxonomy, page builders, multi-channel delivery, or refactoring existing schemas. Trigger when the user mentions schema design, content model, document types, content structure, headless CMS architecture, or asks how to model any content type.
phylogenetics
YuukiAS
Build and analyze phylogenetic trees using MAFFT (multiple alignment), IQ-TREE 2 (maximum likelihood), and FastTree (fast NJ/ML). Visualize with ETE3 or FigTree. For evolutionary analysis, microbial genomics, viral phylodynamics, protein family analysis, and molecular clock studies.