optimize-shiny-performance
About
This skill helps developers profile and optimize slow or unresponsive Shiny applications using tools like profvis, caching (bindCache/memoise), async operations, and ExtendedTask. It's designed for scenarios involving bottlenecks, high concurrent load, or production deployment preparation. The skill provides actionable techniques to improve performance and resource management.
Quick Install
Claude Code
Recommendednpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/optimize-shiny-performanceCopy and paste this command in Claude Code to install this skill
Documentation
Optimize Shiny Performance
Profile, diagnose, and optimize Shiny application performance through caching, async operations, and reactive graph optimization.
When to Use
- Shiny app feels slow or unresponsive during user interaction
- Server resources are exhausted under concurrent user load
- Specific operations (data loading, plotting, computation) create bottlenecks
- Preparing an app for production deployment with many users
Inputs
- Required: Path to the Shiny application
- Required: Description of the performance problem (slow load, laggy interaction, high memory)
- Optional: Number of expected concurrent users
- Optional: Available server resources (RAM, CPU cores)
- Optional: Whether the app uses a database or external API
Procedure
Step 1: Profile the Application
# Profile with profvis
profvis::profvis({
shiny::runApp("path/to/app", display.mode = "normal")
})
# Or profile specific operations
profvis::profvis({
result <- expensive_computation(data)
})
Identify the top bottlenecks:
- Data loading: How long does initial data fetch take?
- Reactive recalculation: Which reactives fire most often?
- Rendering: Which outputs take the longest to render?
- External calls: Database queries, API requests, file I/O?
Use the reactive log for reactive graph analysis:
# Enable reactive logging
options(shiny.reactlog = TRUE)
shiny::runApp("path/to/app")
# Press Ctrl+F3 in the browser to view the reactive graph
Got: Clear identification of the 2-3 biggest bottlenecks.
If fail: If profvis does not show useful detail, wrap specific sections with profvis::profvis(). If reactlog is overwhelming, focus on one interaction at a time.
Step 2: Optimize Reactive Graph
Reduce unnecessary reactive invalidations:
# BAD: Recomputes on ANY input change
output$plot <- renderPlot({
data <- load_data() # Runs every time
filtered <- data[data$category == input$category, ]
plot(filtered)
})
# GOOD: Isolate data loading from filtering
raw_data <- reactive({
load_data()
}) |> bindCache() # Cache the expensive part
filtered_data <- reactive({
raw_data()[raw_data()$category == input$category, ]
})
output$plot <- renderPlot({
plot(filtered_data())
})
Use isolate() to prevent unnecessary invalidations:
# Only recompute when the button is clicked, not on every input change
output$result <- renderText({
input$compute # Take dependency on button
isolate({
paste("N =", input$n, "Mean =", mean(rnorm(input$n)))
})
})
Use debounce() and throttle() for high-frequency inputs:
# Debounce text input — wait 500ms after user stops typing
search_text <- reactive(input$search) |> debounce(500)
# Throttle slider — update at most every 250ms
slider_value <- reactive(input$slider) |> throttle(250)
Got: Reactive graph fires only necessary recalculations.
If fail: If removing a dependency breaks functionality, use req() to add explicit guards instead of relying on implicit reactive dependencies.
Step 3: Implement Caching
bindCache for Shiny Outputs
output$plot <- renderPlot({
create_expensive_plot(filtered_data())
}) |> bindCache(input$category, input$date_range)
output$table <- renderDT({
expensive_query(input$filters)
}) |> bindCache(input$filters)
bindCache uses input values as cache keys. When the same inputs occur again, the cached result is returned immediately.
memoise for Functions
# Cache expensive function results
load_reference_data <- memoise::memoise(
function(dataset_name) {
readr::read_csv(paste0("data/", dataset_name, ".csv"))
},
cache = cachem::cache_disk("cache/", max_age = 3600)
)
App-level Data Pre-computation
# In global.R or outside server function — computed once at app startup
reference_data <- readr::read_csv("data/reference.csv")
model <- readRDS("models/trained_model.rds")
server <- function(input, output, session) {
# reference_data and model are available to all sessions
# without reloading
}
Got: Repeated operations use cached results; response time drops significantly.
If fail: If cache grows too large, set max_age or max_size limits. If cached values are stale, reduce max_age or add a cache-clear button. If bindCache causes errors, ensure cache key inputs are serializable.
Step 4: Add Async for Long Operations
Use ExtendedTask (Shiny >= 1.8.1) for long-running computations:
server <- function(input, output, session) {
# Define the extended task
analysis_task <- ExtendedTask$new(function(data, params) {
promises::future_promise({
# This runs in a background process
run_heavy_analysis(data, params)
})
}) |> bind_task_button("run_analysis")
# Trigger the task
observeEvent(input$run_analysis, {
analysis_task$invoke(dataset(), input$params)
})
# Use the result
output$result <- renderTable({
analysis_task$result()
})
}
For apps on Shiny < 1.8.1, use promises directly:
library(promises)
library(future)
plan(multisession, workers = 4)
server <- function(input, output, session) {
result <- eventReactive(input$compute, {
future_promise({
Sys.sleep(5) # Simulate long computation
expensive_analysis(isolate(input$params))
})
})
output$table <- renderTable({
result()
})
}
Got: Long operations do not block the UI; other users can interact while computation runs.
If fail: If future_promise errors, check that plan(multisession) is set. If variables are not available in the future, pass them explicitly — futures run in separate R processes.
Step 5: Optimize Rendering
Reduce rendering overhead:
# Use plotly for interactive plots instead of re-rendering
output$plot <- plotly::renderPlotly({
plotly::plot_ly(filtered_data(), x = ~x, y = ~y, type = "scatter")
})
# Use server-side DT for large tables
output$table <- DT::renderDataTable({
DT::datatable(large_data(), server = TRUE, options = list(
pageLength = 25,
processing = TRUE
))
})
# Conditional UI to avoid rendering hidden elements
output$details <- renderUI({
req(input$show_details)
expensive_details_ui()
})
Got: Rendering operations are faster and do not block the UI.
If fail: If plotly is slow with large datasets, use toWebGL() for WebGL rendering or downsample data before plotting.
Step 6: Validate Performance Improvements
# Before/after benchmarking
system.time({
shiny::testServer(myModuleServer, args = list(...), {
session$setInputs(category = "A")
session$flushReact()
})
})
# Load testing with shinyloadtest
shinyloadtest::record_session("http://localhost:3838")
shinyloadtest::shinycannon(
"recording.log",
"http://localhost:3838",
workers = 10,
loaded_duration_minutes = 5
)
shinyloadtest::shinyloadtest_report("recording.log")
Got: Measurable improvement in response times and/or concurrent user capacity.
If fail: If performance did not improve, re-profile to find the next bottleneck. Performance optimization is iterative — fix the biggest bottleneck first, then re-measure.
Validation
- Profiling identifies specific bottlenecks (not guessing)
- Reactive graph has no unnecessary invalidation chains
- Expensive operations use caching (bindCache or memoise)
- Long-running computations use async (ExtendedTask or promises)
- High-frequency inputs use debounce/throttle
- Large datasets use server-side processing
- Performance improvement is measurable (before/after timing)
Pitfalls
- Premature optimization: Profile first. The bottleneck is rarely where you think it is.
- Cache invalidation bugs: If users see stale data, the cache key does not include all relevant inputs. Add missing dependencies to
bindCache(). - Future variable scoping:
future_promiseruns in a separate process. Global variables, database connections, and reactive values must be captured explicitly. - Reactive spaghetti: If the reactive graph is too complex to understand, the app needs architectural refactoring (modules), not just caching.
- Over-caching: Caching everything wastes memory. Only cache operations that are expensive AND have repeated input patterns.
Related Skills
build-shiny-module— modular architecture for maintainable reactive codescaffold-shiny-app— choose the right app framework from the startdeploy-shiny-app— deploy optimized apps with appropriate server resourcestest-shiny-app— performance regression tests
GitHub Repository
Related Skills
content-collections
MetaThis skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.
polymarket
MetaThis skill enables developers to build applications with the Polymarket prediction markets platform, including API integration for trading and market data. It also provides real-time data streaming via WebSocket to monitor live trades and market activity. Use it for implementing trading strategies or creating tools that process live market updates.
creating-opencode-plugins
MetaThis skill helps developers create OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It provides the plugin structure, event API specifications, and implementation patterns for JavaScript/TypeScript modules. Use it when you need to intercept, monitor, or extend the OpenCode AI assistant's lifecycle with custom event-driven logic.
sglang
MetaSGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.
