Troubleshoot errors with LLMs

The buggy package provides tools for automatically explaining and fixing R errors using large language models (LLMs). When an error occurs, buggy can analyze the error message, backtrace, and context to provide a human-friendly explanation and suggest fixes.

• buggy_enable(): Attaches a global error handler that captures errors and provides clickable options to explain or fix them.

• buggy_explain(): Explains the most recent error using an LLM, offering detailed context about what went wrong and why.

• buggy_fix(): Attempts to automatically fix the most recent error by generating and applying a code fix. If buggy can find the relevant file lines, it will modify the lines directly. Regardless, it will print the proposed fix out to the console.

Usage

buggy_enable(chat = getOption(".buggy_chat"))

buggy_explain()

buggy_fix()

Arguments

chat

An ellmer Chat object to use for interacting with the language model. If not provided, uses the value from getOption(".buggy_chat"). Set e.g. options(.buggy_chat = ellmer::chat_claude(model = "claude-3-7-sonnet-latest")) in your .Rprofile.

Examples

if (FALSE) { # \dontrun{
# Attach the error handler at the start of your session:
buggy_enable()

# Code that will error:
sum(1, "n")

# If an error occurs, you'll get interactive links to explain or fix
# Alternatively, you can call these functions directly:
buggy_explain()
buggy_fix()
} # }