TDD Cycle

Step 3: GREEN — Minimal implementation

Skill: r-tdd Action: Write the minimum code in R/feature-name.R to make the test pass. Do NOT add logic beyond what the test requires. Run devtools::test_active_file(). Gate: Test runs and PASSES.

Step 4: REFACTOR

Action: Improve code quality — naming, structure, duplication — while keeping tests green. Run devtools::test() to verify no regressions across the package. Gate: All tests pass. No new warnings.

Step 5: REVIEW

Agent: r-code-reviewer (scope: full) Action: Review the new implementation and test files for style, correctness, and performance. Gate: No CRITICAL findings. HIGH findings addressed.

Step 6: COVERAGE

Skill: r-tdd Action: Run covr::package_coverage() and check function-level coverage for the new code. Gate: Coverage >= 80% for new functions.

Abort Conditions

• Test passes on first run in RED phase — the test is wrong, not the code. Rewrite the test before proceeding.
• CRITICAL finding in code review that requires architectural change — stop and re-plan.
• Coverage drops below 80% after refactoring — add tests before continuing.

Examples

Example: Adding a validation function

Prompt: "Add validate_input() that checks a data frame has required columns."

Flow: Setup (create test file) → RED (write test for valid input + missing columns) → GREEN (implement with cli::cli_abort()) → REFACTOR (clean up) → REVIEW (code reviewer) → COVERAGE (verify 80%+)

# RED — test file
test_that("validate_input errors on missing columns", {
  df <- tibble::tibble(id = 1:3)
  expect_error(
    validate_input(df, required = c("id", "value")),
    class = "pkg_missing_columns"
  )
})

# GREEN — minimal implementation
validate_input <- function(df, required) {
  missing <- setdiff(required, names(df))
  if (length(missing) > 0) {
    cli::cli_abort(
      "Column{?s} {.val {missing}} not found in data.",
      class = "pkg_missing_columns"
    )
  }
  invisible(df)
}

Example: Bug fix with regression test

Prompt: "Fix calc_growth() — it returns NA for single-row groups."

Flow: Setup (open existing test file) → RED (write test exposing the bug — single-row group) → GREEN (add guard for n() > 1) → REFACTOR → REVIEW → COVERAGE

# RED — expose the bug
test_that("calc_growth handles single-row groups", {
  df <- tibble::tibble(group = "A", date = as.Date("2024-01-01"), value = 100)
  result <- calc_growth(df)
  expect_true(is.na(result$growth[1]))  # expected NA, not error
})