MLOps
模型部署、评估与运维
story-outliner
GongLingRui
阅读理解故事文本，总结人物、关系、情节，整理成流畅大纲。适用于快速了解故事核心、为剧本创作提供大纲基础
story-outline-evaluator
GongLingRui
专业评估故事大纲，从市场潜力、创新属性、内容亮点维度判断评分。适用于故事大纲质量评估、IP改编潜力判断、项目立项决策
story-summarizer
GongLingRui
基于故事文本提炼主要情节与要点，生成完整故事梗概。适用于快速了解故事内容、剧本改编、项目推介
drama-planner
GongLingRui
制定竖屏短剧专业策划方案，含情绪价值分析、黄金三秒钩子设计、三幕式结构规划等。适用于短剧项目前期策划、商业化方案设计、创作指导
plot-keypoints
GongLingRui
梳理故事主线，提炼并按发展阶段排列主要情节点。适用于快速掌握故事结构、制作大纲及剧本改编结构梳理
plot-points-analyzer
GongLingRui
分析故事情节点，识别关键情节与转折点。适用于深度分析情节结构、评估情节发展有效性
advanced-evaluation
shipshitdev
Master LLM-as-a-Judge evaluation techniques including direct scoring, pairwise comparison, rubric generation, and bias mitigation. Use when building evaluation systems, comparing model outputs, or establishing quality standards for AI-generated content.
check-context-communication
dykyi-roman
Audits Bounded Context communication patterns. Checks Context Map relationships (Shared Kernel, ACL, Open Host), event vs direct calls, and anti-corruption layer usage.
cats-mtl-typed-errors
alexandru
Scala typed errors with Cats MTL Raise/Handle and allow/rescue. Use for designing custom domain error types without EitherT, while keeping Cats Effect and ecosystem composition. Covers Scala 2/3 syntax and IO-only or F[_] usage.
claude-usage
tdimino
Report ground-truth Claude token consumption and estimated cost by parsing JSONL session files directly. Use when checking API spend, auditing token usage by project/session/model, generating daily/weekly/monthly cost reports, or diagnosing ccusage undercounting. Includes subagent files that ccusage ignores.
parakeet
tdimino
Local speech-to-text via Handy app (push-to-talk) and NeMo CLI scripts. Parakeet V3: 25 languages, auto-detection, ~30x realtime on M4 Max, 6% WER. This skill should be used when transcribing audio files or dictating voice input.
llama-cpp
tdimino
Secondary local LLM inference engine via llama.cpp. This skill should be used when running GGUF models directly, loading LoRA adapters for Kothar, benchmarking inference speed, or serving models via llama-server. Complements Ollama (which remains primary for RLAMA and general use).
codex-cto
tdimino
Inverted orchestration—Codex CLI acts as CTO (planning and reviewing), Claude Code executes with native tools. Use when a task benefits from external architectural oversight before and after implementation.
smolvlm
tdimino
Local vision-language model for image analysis using SmolVLM-2B
Behavioral Modification
pjt222
Agent Almanac — a curated reference of executable skills, specialist agents, and teams for AI-assisted development, with interactive visualization
Causal Inference
aj-geddes
Determine cause-and-effect relationships using propensity scoring, instrumental variables, and causal graphs for policy evaluation and treatment effects
capacity-planning
aj-geddes
Analyze team capacity, plan resource allocation, and balance workload across projects. Forecast staffing needs and optimize team utilization while maintaining sustainable pace.
ml-experiment-tracker
openclaw
Plan reproducible ML experiment runs with explicit parameters, metrics, and artifacts. Use before model training to standardize tracking-ready experiment definitions.
ml-model-eval-benchmark
openclaw
Compare model candidates using weighted metrics and deterministic ranking outputs. Use for benchmark leaderboards and model promotion decisions.
dl-transformer-finetune
openclaw
Build transformer fine-tuning run plans with task settings, hyperparameters, and model-card outputs. Use for repeatable Hugging Face or PyTorch finetuning workflows.
gemini
jdrhyne
Use when the user asks to run Gemini CLI for code review, plan review, or big context (>200k) processing. Ideal for comprehensive analysis requiring large context windows. Uses Gemini 3 Pro by default for state-of-the-art reasoning and coding.
codex
jdrhyne
Use when the user asks to run Codex CLI (codex exec, codex resume) or references OpenAI Codex for code analysis, refactoring, or automated editing. Uses GPT-5.2 by default for state-of-the-art software engineering.
agentic-kaggle-skill
FrankS-IntelLab
Kaggle-first end-to-end competition workflow for scored submissions. Use when Codex must run Kaggle or competitive ML workflows through scored submission, including code competitions, validation, metrics, policy-safe public notebook/discussion intel, tabular/text/image modeling, tuning, ensembling/stacking, proactive multi-notebook architectures, producer notebooks that train models and export private Kaggle artifact datasets, downstream consumer notebooks, Kaggle GPU offload, kagglehub access, hidden-test debugging, and public score retrieval.
megatron-memory-estimator
yzlnew
Estimate GPU memory usage for Megatron-based MoE (Mixture of Experts) and dense models. Use when users need to (1) estimate memory from HuggingFace model configs (DeepSeek-V3, Qwen, etc.), (2) plan GPU resource allocation for training, (3) compare different parallelism strategies (TP/PP/EP/CP), (4) determine if a model fits in available GPU memory, or (5) optimize training configurations for memory efficiency.