r-claude-skills

Expert Keras3 deep learning in R with multi-backend support. Use when working with keras3, mentions "keras3", "keras3 em R", "keras3 for R", "Functional API", "custom keras layer", "keras preprocessing", "keras application", "keras subclassing", "JAX backend", "multi-backend keras", "keras3 audio", "keras3 NLP", "keras3 vision", "preprocessing layer", "custom training loop", "model subclassing", "Sequential API", "layer_*", "optimizer_adam", "compile model", "fit model", or discusses deep learning with keras3 in R.

Machine learning paradigm selection guide covering self-supervised, few-shot, weak supervision, transfer learning and meta-learning. Use when mentions "self-supervised learning", "SSL", "few-shot learning", "FSL", "few shot", "weak supervision", "weakly supervised", "limited labeled data", "limited labels", "learning paradigms", "paradigmas de aprendizado", "meta-learning", "transfer learning", "quando usar SSL", "quando usar few-shot", "which learning approach", "escolher paradigma", "choose learning paradigm", "data scarcity", "escassez de dados", "unlabeled data", "dados não rotulados", or asks about learning strategy selection for data-limited scenarios.

Expert Quarto document creation and publishing with R. Use when working with Quarto, mentions "criar relatório", "quarto report", "quarto dashboard", "quarto presentation", "apresentação quarto", "quarto slides", "quarto website", "site quarto", "blog quarto", "quarto book", "livro quarto", "documento quarto", "render quarto", "renderizar quarto", "publicar quarto", "publish quarto", ".qmd file", "arquivo qmd", discusses reproducible research, data storytelling, interactive reports, scientific writing with R, or asks about creating reports, dashboards, presentations, websites, or books with R.

Multi-label audio classification in R using torch for bioacoustics and ecological monitoring. Use when mentions "multi-label audio", "multilabel audio", "multi-label classification", "classificação multi-rótulo", "mel-spectrogram", "mel spectrogram", "audio augmentation", "time-frequency masking", "BCEWithLogitsLoss", "audio preprocessing", "bioacoustics", "bioacústica", "acoustic monitoring", "monitoramento acústico", "sound classification", "classificação de sons", "overlapping species", "espécies sobrepostas", "soundscape analysis", "análise de paisagem sonora", "PAM", "passive acoustic monitoring", "audio multi-label", "torch audio", "torchaudio in R", or building sound classification models with overlapping classes in ecological contexts.

Expert bioacoustic analysis in R using tuneR, seewave, warbleR, bioacoustics, ohun, and soundecology. Use when mentions "análise de áudio em R", "audio analysis in R", "bioacoustics", "bioacústica", "acoustic analysis", "análise acústica", "espectrograma", "spectrogram", "MFCC", "mel-spectrogram", "detecção de eventos sonoros", "event detection", "sound detection", "passive acoustic monitoring", "PAM", "monitoramento acústico passivo", "índices ecoacústicos", "ecoacoustic indices", "acoustic indices", "tuneR", "seewave", "warbleR", "bioacoustics package", "ohun", "soundecology", "feature extraction from audio", "extração de features de áudio", "classificação de sons", "sound classification", "bird sound", "som de aves", "animal sound", "análise de vocalizações", "vocalization analysis", "acoustic ecology", "ecologia acústica", or working with environmental audio, animal vocalizations, or soundscape analysis.

Expert R with Databricks and Apache Spark using sparklyr. Use when mentions "databricks", "sparklyr", "spark com R", "Spark in R", "spark with R", "distributed data", "big data em R", "big data in R", "cluster computing", "databricks connect", "spark_connect", "remote spark", "delta lake", "delta table", "lakehouse", "sdf_", "ml_* functions", "spark ML", "escalar análise", "scale data science", "distributed computing", "parallel processing with spark", "DBR", "databricks runtime", "unity catalog", "broadcast join", "spark dataframe", "repartition", "shuffle", "executor", "driver node", or working with large-scale data analysis in R on Databricks. ONLY R - do NOT activate for Python, PySpark, Scala Spark, or other non-R big data tools.

Expert R data science using tidyverse and tidymodels. Use when working with data in R, mentions "tidyverse", "dplyr", "data wrangling", "análise de dados", "análise de dados em R", "data analysis", "data analysis in R", "análise estatística", "statistical analysis", "estatística em R", "machine learning", "machine learning in R", "ML em R", "aprendizado de máquina", "statistical modeling", "modelagem estatística", "ggplot2", "data visualization", "visualização de dados", "predictive modeling", "modelagem preditiva", "feature engineering", "engenharia de features", "model training", "treinar modelo", "treinamento de modelo", "cross-validation", "validação cruzada", "ciência de dados", "data science", "explorar dados", "explore data", "EDA", "análise exploratória", "exploratory analysis", or any data science task in R. ONLY activate for R language - do NOT activate for Python, pandas, scikit-learn, numpy, or other non-R data science tools.

Deep learning in R using torch and keras3 for neural networks across domains. Use when mentions "deep learning em R", "deep learning in R", "torch", "torch em R", "torch for R", "keras3", "keras3 em R", "neural network", "rede neural", "redes neurais", "CNN", "convolutional neural network", "rede convolucional", "RNN", "LSTM", "GRU", "recurrent neural network", "transformer", "GPU em R", "GPU in R", "train neural network", "treinar rede neural", "image classification", "classificação de imagens", "text classification", "classificação de texto", "sound classification", "audio classification", "classificação de sons", "audio deep learning", "deep learning com áudio", "spectrogram CNN", "CRNN", "mel-spectrogram", "torchaudio", "luz package", "build neural network", "construir rede neural", "transfer learning", "fine-tuning", "pretrained model", "modelo pré-treinado", or building/training neural network models in R for vision, NLP, audio, time series, or tabular data.

Expert interactive data visualization in R with plotly - web-based graphics, animations, Shiny integration, and 3D plots. Use when user works with plotly, mentions "plotly", "plot_ly", "ggplotly", "interactive plot", "interactive visualization", "visualização interativa", "gráfico interativo", "animated plot", "animação", "animation", "plotly em R", "plotly for R", "plotly in R", "web graphics", "HTML widget", "3D plot", "3D surface", "choropleth", "mapbox", "plotly animation", "plotly shiny", "event_data", "plotlyOutput", "renderPlotly", "hover template", "plotly trace", "add_trace", "layout plotly", or discusses interactive/web-based visualization in R. Do NOT activate for static ggplot2 plots without plotly conversion.

Expert TensorFlow for R using tensorflow package and keras3 integration. Use when working with TensorFlow in R, mentions "tensorflow em R", "tensorflow for R", "tensorflow R interface", "install_tensorflow", "TensorFlow deployment", "TensorFlow infrastructure", "keras3 backend", "TensorFlow graph", "SavedModel", "tf_function", discusses TensorFlow-specific features, deployment pipelines, or integration with Python TensorFlow code. ONLY R - do NOT activate for Python TensorFlow.

Expert R data science using tidymodels for machine learning. Use when working with tidymodels, recipes, parsnip, tune, workflows, user mentions "machine learning in R", "ML em R", "predictive modeling", "modelagem preditiva", "modelo de classificação", "classification model", "modelo de regressão", "regression model", "feature engineering", "engenharia de features", "model tuning", "tunear modelo", "tuning", "tunar", "tune boosting", "tune gradient boosting", "xgboost tuning", "random forest tuning", "hyperparameters", "hiperparâmetros", "cross-validation", "validação cruzada", "cv", "recipe steps", "data preprocessing", discusses ML workflows, or model development in R.

Master torch (PyTorch for R) deep learning with maximum flexibility and control. Use when mentions "torch", "torch em R", "torch for R", "torch in R", "PyTorch R", "pt", "nn_module", "nn_linear", "nn_conv", "nn_sequential", "torch_tensor", "autograd", "automatic differentiation", "custom training loop", "manual training", "backpropagation", "gradient manipulation", "research model", "custom loss function", "custom layer", "torch dataset", "dataloader", "GPU torch", "CUDA torch", "treinar com torch", "rede neural torch", "CNN torch", "RNN torch", "LSTM torch", "GRU torch", "attention mechanism", "transformer torch", "audio torch", "NLP torch", "vision torch", "time series torch", "torch flexibility", "torch control", "low-level training", "research deep learning", "experimental models", "dynamic computation graph", "when use torch", "torch vs keras", "torch or keras3", "torch benefits", "LibTorch", "mlverse torch", "torch for research", "torch for production", or working with torch R package for deep learning,

Advanced feature engineering and selection strategies in R. Use when mentions "feature engineering strategies", "feature selection", "categorical encoding methods", "likelihood encoding", "target encoding", "entity embeddings", "feature hashing", "interaction detection", "systematic feature selection", "wrapper methods", "filter methods", "greedy search", "global search", "genetic algorithms", "Box-Cox transformation", "quando usar cada encoding", "when to use each encoding", "escolher método de encoding", "choose encoding method", "detectar interações", "detect interactions", "selecionar features", "select features", "feature selection methods", "escolher entre encodings", "compare encoding methods", "which encoding to use", "qual encoding usar", or discusses strategic decisions about preprocessing beyond basic recipe implementation.

Patterns for Bayesian inference in R using brms, including multilevel models, DAG validation, and marginal effects. Use when mentions "Bayesian in R", "Bayesian with R", "brms", "Stan", "cmdstanr", "multilevel model", "hierarchical model", "random effects", "prior specification", "posterior distribution", "MCMC", "Markov Chain Monte Carlo", "DAG", "causal inference", "marginal effects", "tidybayes", or performing Bayesian statistical analysis in R. ONLY R - do NOT activate for Python Bayesian (PyMC3, PyMC, Pyro), NOT for Julia (Turing.jl).

Expert text mining and NLP in R using tidytext and textrecipes. Use when analyzing text data, mentions "text analysis in R", "análise de texto em R", "NLP in R", "processamento de linguagem natural em R", "sentiment analysis", "análise de sentimento", "topic modeling", "modelagem de tópicos", "tidytext", "textrecipes", "tokenization", "tokenização", "tokenizar", "TF-IDF", "text classification", "classificação de texto", "word embeddings", "n-gram", "ngram", "natural language processing", "analyze reviews", "customer reviews", "analyze text", "preprocess text", "text preprocessing", "preprocessar texto", or any text/NLP task in R. ONLY R - do NOT activate for Python NLP (spaCy, NLTK, TextBlob), NOT for general NLP without R context.

Expert R data manipulation with tidyverse - dplyr, tidyr, purrr, stringr, forcats, lubridate. Use when working with tidyverse, mentions "filter", "select", "mutate", "summarize", "summarise", "arrange", "group_by", "join", "joins", "dplyr verbs", "data wrangling", "manipulação de dados", "data manipulation", "tidyr pivoting", "pivot_longer", "pivot_wider", "pivot", "purrr map", "map", "map_dbl", "purrr", "string manipulation", "manipulação de strings", "stringr", "str_detect", "str_replace", "regex", "factors", "forcats", "fct_reorder", "fct_lump", "fct_collapse", "fct_infreq", "fct_recode", "reorder factors", "reordenar fatores", "reordenar níveis", "factor levels", "níveis de fator", "collapse levels", "combinar níveis", "agrupar níveis", "dates in R", "datas em R", "lubridate", "ymd", "mdy", "dmy", "parse dates", "parsear datas", "parse date", "parsing dates", "year()", "month()", "day()", "hour()", "minute()", "date arithmetic", "aritmética de datas", "date math", "floor_date", "ceiling_date", "round_date

R object-oriented programming guide for S7, S3, S4, and vctrs. Use when mentions "orientado a objetos", "orientado a objetos em R", "object-oriented", "object-oriented in R", "OOP", "OOP in R", "OOP em R", "POO", "POO em R", "classes", "classes em R", "classes in R", "métodos", "métodos em R", "methods", "methods in R", "S3 class", "S4 class", "S7 class", "S7", "classe S3", "classe S4", "classe S7", "sistema S3", "sistema S4", "sistema S7", "S3 system", "S4 system", "S7 system", "method dispatch", "despacho de métodos", "dispatch de métodos", "generic functions", "funções genéricas", "generics", "genéricos", "inheritance", "herança", "vctrs", "vctrs package", "methods", "methods package", "setClass", "setGeneric", "setMethod", "new_class", "new_generic", "new_property", "criar classe", "create class", "design class", "desenhar classe", "definir classe", "define class", "sistema de objetos", "object system", "class system", "sistema de classes", "definir métodos", "define methods", "criar genéricos", "create g

Expert time series forecasting and analysis in R using fable/tsibble/feasts. ONLY R - do NOT activate for Python time series (statsmodels, prophet in Python). Use when user mentions "série temporal", "time series", "time series in R", "time series with R", "previsão", "forecasting", "forecast in R", "predict time series", "prever série temporal", "analisar série temporal", "analyze time series", "temporal data in R", "dados temporais", "time-based prediction", "ARIMA in R", "ARIMA model", "ETS in R", "ETS model", "fable", "tsibble", "feasts", "forecast package", "prophet in R", "sazonalidade", "seasonality", "seasonal patterns", "padrões sazonais", "lidar com sazonalidade", "handle seasonality", "tendência", "trend", "decomposição", "decomposition", "temporal", "time-based", "temporal patterns", "padrões temporais", or any R-specific time series task.

Modern tidyverse patterns for R including pipes, joins, grouping, purrr, and stringr. Use when mentions "native pipe", "pipe nativo", "pipe operator", "operador pipe", "|>", "new pipe", "novo pipe", "is this the new pipe", "é o novo pipe", "join_by", "join_by()", "join syntax", "sintaxe de join", "relacionamento de join", ".by grouping", ".by", "per-operation grouping", "agrupamento por operação", "agrupamento por operacao", "across()", "pick()", "reframe()", "list_rbind", "list_rbind()", "list_cbind", "list_cbind()", "modern tidyverse", "tidyverse moderno", "latest tidyverse", "tidyverse atual", "dplyr 1.1", "dplyr 1.1 features", "recursos dplyr 1.1", "latest dplyr", "modern patterns", "padrões modernos", "padroes modernos", "usar pipe nativo", "use native pipe", "usar |>", "use |>", "agrupar por operação", "agrupar por operacao", "per-operation group", writing modern tidyverse R code with latest patterns, or implementing current dplyr 1.1+ features. ONLY FOR R - do NOT activate for pandas pipe, Python dplyr

Relational data modeling with {dm} package - create, visualize, and manipulate multi-table data models with primary/foreign keys. Use when mentions "dm package", "pacote dm", "relational data", "dados relacionais", "primary key", "chave primária", "foreign key", "chave estrangeira", "data model", "modelo de dados", "multi-table", "múltiplas tabelas", "tabelas relacionadas", "relate tables", "relacionar tabelas", "database schema", "esquema de banco", "esquema de dados", "dm_from_data_frames", "dm_add_pk", "dm_add_fk", "dm_draw", "dm_flatten", "dm_filter", "dm_zoom_to", "visualize schema", "visualizar esquema", "create data model", "criar modelo", "model relational data", "modelar dados relacionais", "work with related data", "trabalhar com dados relacionados", "multiple related tables", "várias tabelas relacionadas", or working with related data frames, relational databases, or referential integrity in R.

R package development guide covering dependencies, API design, testing, and documentation. Use when mentions "R package", "pacote R", "pacote r", "devtools", "usethis", "roxygen2", "testthat", "pkgdown", "criar pacote", "create package", "desenvolver pacote", "develop package", "package development", "desenvolvimento de pacote", "DESCRIPTION", "NAMESPACE", "CRAN submission", "submissão CRAN", "submissão ao CRAN", "package documentation", "documentação de pacote", "package structure", "estrutura de pacote", "vignette", "vinheta", "vignet", "R CMD check", "check package", "verificar pacote", "build package", "construir pacote", "criar um pacote R", "make an R package", "building R packages", "construindo pacotes R", or developing R packages. ONLY R packages - do NOT activate for Python packages, npm packages, pip, setuptools, or JavaScript packages.

R style guide covering naming conventions, spacing, layout, and function design best practices. Use when mentions "estilo de código", "code style", "formatação", "formatting", "format code", "formatar código", "convenções", "conventions", "naming convention", "convenção de nomes", "snake_case", "camelCase", "styler", "lintr", "apply styler", "run styler", "aplicar styler", "executar styler", "tidyverse style", "estilo tidyverse", "R coding standards", "padrões de código R", "padrões de código", "best practices", "boas práticas", "code quality", "qualidade de código", "check style", "verificar estilo", "R style guide", "guia de estilo R", "function design", "design de funções", "clean code R", "código limpo", "indentation", "indentação", "spacing", "espaçamento", "assignment operator", "operador de atribuição", "arrow operator", "naming variables", "nomear variáveis", "naming functions", "nomear funções", "code organization", "organização de código", "file structure", "estrutura de arquivos", "comments", "come

rlang metaprogramming patterns for data-masking, injection operators, and dynamic dots. Use when mentions "tidy evaluation", "avaliação tidy", "rlang", "metaprogramming", "metaprogramação", "metaprogramação em R", "data-masking", "data masking", "mascaramento de dados", "embrace", "embrace operator", "operador embrace", "{{}}", "enquo", "!!", "!!!", "injection", "injeção", "inject", "injetar", "dynamic dots", "...", "NSE", "non-standard evaluation", "avaliação não padrão", "avaliação não-padrão", "quo", "quos", "sym", "syms", "data_sym", "data_syms", ".data", ".env", "pronouns", "pronomes", "defuse", "defusing", "write functions with tidy eval", "escrever funções com tidy eval", "funções com tidy eval", "usar enquo", "use enquo", "usar embrace", "use embrace", "usar !!", "use !!", "usar !!!", "use !!!", "forward arguments", "encaminhar argumentos", or writing functions that use tidy evaluation in R.

Expert ggplot2 data visualization in R - grammar of graphics, geoms, themes, scales, faceting, and styling. Use when user works with ggplot2, mentions "ggplot", "geom_", creates visualizations in R, asks about plot customization, "customizar theme", "customize theme", "customizar plot", themes, "facet_wrap", "facet_grid", "faceting", "facetas", "anotações", "annotations", "annotate", "adicionar anotações", "color scale", "scales", "escala de cores", "gráfico", "plot", "visualização", or data visualization best practices.

R performance best practices including profiling, benchmarking, vctrs, and optimization strategies. Use when mentions "profiling", "profvis", "benchmark", "bench::mark", "slow code", "código lento", "lento", "slow", "optimize R", "otimizar R", "otimizar código", "otimizar", "optimize", "optimize code", "vectorization", "vectorizar", "vectorize", "performance", "memory usage", "bottleneck", "gargalo", "speed up", "acelerar", "parallel processing", "Rcpp", "system.time", or optimizing R code performance. ONLY R - do NOT activate for Python, C++, JavaScript optimization.

Expert Shiny app development in R - reactive programming, UI design, modules, performance, and security. Use when working with shiny, shinydashboard, or building interactive R web applications. Triggers on "dashboard interativo", "interactive dashboard", "dashboard em R", "R dashboard", "aplicativo R", "R app", "aplicação Shiny", "Shiny app", shiny library imports, server/ui function patterns like "ui <- fluidPage", "server <- function", reactive expressions, renderPlot, renderTable, observeEvent, reactiveVal, or user requests for Shiny help, web applications in R, or reactive programming guidance.

Test-driven development workflow for R using testthat. Use when mentions "TDD", "test-driven", "test driven development", "testthat", "test_that", "expect_", "unit test", "unit testing", "testes unitários", "test coverage", "cobertura de testes", "covr", "red-green-refactor", "test first", "write tests", "escrever testes", or writing new features, fixing bugs, or refactoring code. Enforces test-first development with 80%+ coverage.

Create new Claude Code skills following best practices. Use when user asks to "create a skill", "criar skill", "criar um skill", "make a new skill", "build a skill", "generate skill", "gerar skill", "novo skill", "new skill", "custom skill", "skill personalizado", mentions "skill maker", "skillMaker", "skill generator", "gerador de skill", discusses "creating Claude Code automations", "criar automação", "build automation", "skill development", "desenvolvimento de skill", "skill creation", "criação de skill", wants help "building custom Claude skills", "construir skills", "design a skill", "desenhar skill", or asks about "how to make a skill", "como criar skill", "skill template", "template de skill", "skill structure", "estrutura de skill", "skill patterns", "padrões de skill".