---

name: karthik-r-code-style description: "Karthik's R code layout: one pipe per line, right-assigned pipelines, expanded multi-arg calls, +-per-line ggplot, 2-space indent."

Karthik R Code Style

Use whenever you write or edit R for Karthik - scripts, .Rmd/.qmd chunks, ggplot code, anything. This skill governs layout: line breaks, indentation, bracket expansion, assignment direction, spacing. It does not govern which idioms or which analysis to write - that is karthik-r-analysis-style. Apply both together when generating a notebook; apply this one alone when the request is just "format this" or "write this in my R style".

LLM-generated R packs everything onto long horizontal lines. Karthik's R breaks vertically so each step reads as its own line and a pipeline can be run line by line while exploring. Default to the expanded layout below; don't wait to be asked for it.

The rules

1. One pipe per line

Every %>% or |> ends its line. The source object sits at the pipeline's base indent; each following verb goes on a new line, indented one step (2 spaces) under it.

movie_shows %>%
  arrange(show_id, desc(start_at)) %>%
  distinct(show_id, .keep_all = TRUE) %>%
  mutate(show_date = as_date(start_at))

Never chain two verbs on one line (x %>% filter(...) %>% select(...) on a single line is wrong for anything beyond a throwaway one-liner).

2. Right-assign pipelines

A pipeline that builds a named object ends with right assignment, so the chain can be selected and run top-down. Always put -> at the end of the expression and the target object on the next line, indented 2 spaces from the pipeline's base indent. Never put the target on the same line as ->, even for a short expression.

shows %>%
  summarise(
    first_show_date = min(show_date),
    .by = movie_id
  ) ->
  movie_show_dates

movie_show_dates %>%
  left_join(selected_releases, by = "movie_id", relationship = "one-to-one") ->
  movie_releases

Reserve <- for short, single-line assignments that are not pipelines (outdir <- "report_assets", emph <- "#C6462F"). Reading a file into a name also uses right assignment with the target on the next line:

read_parquet(path) ->
  movie_shows

3. Expand multi-argument calls

When a call carries several arguments, or won't sit comfortably on one line, put ( at the end of the line, one argument (or one logical group) per line indented one step, and the closing ) on its own line at the call's base indent. The next %>% or -> follows the ); after ->, put the target object on the next line.

screens_raw %>%
  summarise(
    rows = n(),
    seat_values = n_distinct(seats),
    .by = screen_id
  ) ->
  screen_capacity

For per-operation grouping (the default in karthik-r-analysis-style), put .by = key or .by = c(key1, key2) last, on its own argument line inside mutate(), summarise(), or another supporting verb. Avoid a surrounding group_by() / ungroup() pair for a single grouped operation.

Don't expand trivial calls. n(), select(chat_id, year), count(uuid) stay inline - the rule is for calls with real argument lists, not every set of parentheses.

4. ggplot: + per line

Each layer ends with +; the next geom_*, scale_*, labs(), annotate(), theme() goes on its own line indented one step under ggplot(...). A single layer with many arguments expands by rule 3.

ggplot(arc, aes(year)) +
  geom_line(aes(y = affection), colour = emph, linewidth = 1.3) +
  geom_point(aes(y = affection), colour = emph, size = 2.4) +
  labs(
    title = "The relationship, in two lines",
    subtitle = "Mean coded affection and tension per chat, by year"
  ) +
  base_theme

5. Indent is 2 spaces, and it nests

One step is two spaces. A bracket opened inside an already-indented line adds another two. Align the closing bracket with the line that opened it.

6. Spacing

Space after every comma. Spaces around = in named arguments and around operators (year >= 2008, by = "movie_id"). This is the modern-notebook convention; ignore the cramped no-space style of the pre-2016 base-R files on disk.

What this is not

Not a linter to run over Karthik's old code - old hand-written files keep their own style, including base-R Allman braces and cramped spacing. This skill shapes new R you generate. Idiom choice (tidyverse over base R, cummean over index loops, no raw SQL) and notebook structure live in karthik-r-analysis-style; final chart design lives in karthik-data-visualization.