4 Inline code completion

Software requirements: Positron or VS Code with R or Python and a GitHub Copilot license + extension (or RStudio with Copilot enabled — see Setup). Alternatively, a local LLM with the Continue extension or similar - see local models.

Inline completion is the gentlest introduction to AI-assisted coding, and probably how you’re already using it if you’ve dabbled at all. You start typing, the assistant guesses what you meant to type next, and you hit tab to accept.

GitHub Copilot calls itself an “AI programming assistant” or “AI pair programmer.” I’ll refer to it more generally as an LLM coding assistant, or just an Assistant — a term I’ll also use for ChatGPT, Claude, Cline and others, since the field moves too fast to have settled on a standard name for this kind of software.

An Assistant adds a layer between you and the LLM. It’s quietly deciding what context to send, managing your context window, and often saving on cost — usually without telling you the system message, temperature, or token limits it’s using. Keep that in mind: you’re not talking directly to the model, you’re talking through some software that’s making its own decisions about what the model sees.

Let’s read in the benthic coral cover and fish count data we’ll use throughout this book:

library(tidyverse)
library(readr)

dat <- read_csv(url(
    "https://raw.githubusercontent.com/cbrown5/example-ecological-data/refs/heads/main/data/benthic-reefs-and-fish/fish-coral-cover-sites.csv"
))

head(dat)
summary(dat)

More details on this data are available here.

Now start a new line and try create a ggplot of secchi (a measure of water clarity — higher values mean clearer water) and pres.topa (count of topa, the bumphead parrotfish). Start typing gg and see what happens.

You should get a recommendation for a ggplot. But it won’t know the variable names, because completion is guessing from your script (and any other open scripts), not from the data itself.

Sometimes the assistant gets stuck in a loop and keeps recommending the same line. To break it out of the loop try typing something new.

Challenge

Clear a new script, load dat, and try tab-completing a ggplot of CB_cover against dist_to_logging_km without typing any variable names yourself first. Note how many attempts it takes before the completion uses the correct column names.