Agent-almanac fail-early-pattern

install
source · Clone the upstream repo
git clone https://github.com/pjt222/agent-almanac
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/pjt222/agent-almanac "$T" && mkdir -p ~/.claude/skills && cp -r "$T/i18n/wenyan-lite/skills/fail-early-pattern" ~/.claude/skills/pjt222-agent-almanac-fail-early-pattern-0f7e18 && rm -rf "$T"
manifest: i18n/wenyan-lite/skills/fail-early-pattern/SKILL.md
source content

Fail Early

If something is going to fail, it should fail as early as possible, as loudly as possible, with as much context as possible. This skill codifies the fail-early pattern: validating inputs at system boundaries, using guard clauses to reject bad state before it propagates, and writing error messages that answer what failed, where, why, and how to fix it.

When to Use

  • Writing or reviewing functions that accept external input (user data, API responses, file contents)
  • Adding input validation to package functions before CRAN submission
  • Refactoring code that silently produces wrong results instead of erroring
  • Reviewing pull requests for error-handling quality
  • Hardening internal APIs against invalid arguments

Inputs

  • Required: Function or module to apply the pattern to
  • Required: Identification of trust boundaries (where external data enters)
  • Optional: Existing error-handling code to refactor
  • Optional: Target language (default: R; also applies to Python, TypeScript, Rust)

Procedure

Step 1: Identify Trust Boundaries

Map where external data enters the system. These are the points that need validation:

  • Public API functions (exported functions in an R package)
  • User-facing parameters
  • File I/O (reading configs, data files, user uploads)
  • Network responses (API calls, database queries)
  • Environment variables and system configuration

Internal helper functions called only by your own validated code generally do not need redundant validation.

Expected: A list of entry points where untrusted data crosses into your code.

On failure: If boundaries are unclear, trace backwards from errors in logs or bug reports to find where bad data first entered.

Step 2: Add Guard Clauses at Entry Points

Validate inputs at the top of each public function, before any work begins.

R (base):

calculate_summary <- function(data, method = c("mean", "median", "trim"), trim_pct = 0.1) {
  # Guard: type check
  if (!is.data.frame(data)) {
    stop("'data' must be a data frame, not ", class(data)[[1]], call. = FALSE)
  }
  # Guard: non-empty
  if (nrow(data) == 0L) {
    stop("'data' must have at least one row", call. = FALSE)
  }
  # Guard: argument matching
  method <- match.arg(method)
  # Guard: range check
  if (!is.numeric(trim_pct) || trim_pct < 0 || trim_pct > 0.5) {
    stop("'trim_pct' must be a number between 0 and 0.5, got: ", trim_pct, call. = FALSE)
  }
  # --- All guards passed, begin real work ---
  # ...
}

R (rlang/cli — preferred for packages):

calculate_summary <- function(data, method = c("mean", "median", "trim"), trim_pct = 0.1) {
  rlang::check_required(data)
  if (!is.data.frame(data)) {
    cli::cli_abort("{.arg data} must be a data frame, not {.cls {class(data)}}.")
  }
  if (nrow(data) == 0L) {
    cli::cli_abort("{.arg data} must have at least one row.")
  }
  method <- rlang::arg_match(method)
  if (!is.numeric(trim_pct) || trim_pct < 0 || trim_pct > 0.5) {
    cli::cli_abort("{.arg trim_pct} must be between 0 and 0.5, not {.val {trim_pct}}.")
  }
  # ...
}

General (TypeScript):

function calculateSummary(data: DataFrame, method: Method, trimPct: number): Summary {
  if (data.rows.length === 0) {
    throw new Error(`data must have at least one row`);
  }
  if (trimPct < 0 || trimPct > 0.5) {
    throw new RangeError(`trimPct must be between 0 and 0.5, got: ${trimPct}`);
  }
  // ...
}

Expected: Every public function opens with guard clauses that reject invalid input before any side effects or computation.

On failure: If validation logic is getting long (>15 lines of guards), extract a

validate_*
helper or use
stopifnot()
for simple type assertions.

Step 3: Write Meaningful Error Messages

Every error message should answer four questions:

  1. What failed — which parameter or operation
  2. Where — function name or context (automatic with
    cli::cli_abort
    )
  3. Why — what was expected vs. what was received
  4. How to fix — when the fix is non-obvious

Good messages:

# What + Why (expected vs. actual)
stop("'n' must be a positive integer, got: ", n, call. = FALSE)

# What + Why + How to fix
cli::cli_abort(c(
  "{.arg config_path} does not exist: {.file {config_path}}",
  "i" = "Create it with {.run create_config({.file {config_path}})}."
))

# What + context
cli::cli_abort(c(
  "Column {.val {col_name}} not found in {.arg data}.",
  "i" = "Available columns: {.val {names(data)}}"
))

Bad messages:

stop("Error")                    # What failed? No idea
stop("Invalid input")           # Which input? What's wrong with it?
stop(paste("Error in step", i)) # No actionable information

Expected: Error messages are self-documenting — a developer seeing the error for the first time can diagnose and fix it without reading source code.

On failure: Review the three most recent bug reports. If any required reading source code to understand, their error messages need improvement.

Step 4: Prefer stop() Over warning()

Use

stop()
(or
cli::cli_abort()
) when the function cannot produce a correct result. Use
warning()
only when the function can still produce a meaningful result but the caller should know about a concern.

Rule of thumb: If a user could silently get a wrong answer, that is a

stop()
, not a
warning()
.

# CORRECT: stop when result would be wrong
read_config <- function(path) {
  if (!file.exists(path)) {
    stop("Config file not found: ", path, call. = FALSE)
  }
  yaml::read_yaml(path)
}

# CORRECT: warn when result is still usable
summarize_data <- function(data) {
  if (any(is.na(data$value))) {
    warning(sum(is.na(data$value)), " NA values dropped from 'value' column", call. = FALSE)
    data <- data[!is.na(data$value), ]
  }
  # proceed with valid data
}

Expected:

stop()
is used for conditions that would produce incorrect results;
warning()
is reserved for degraded-but-valid outcomes.

On failure: Audit existing

warning()
calls. If the function returns nonsense after the warning, change it to
stop()
.

Step 5: Use Assertions for Internal Invariants

For conditions that "should never happen" in correct code, use assertions. These catch programmer errors during development:

# R: stopifnot for internal invariants
process_chunk <- function(chunk, total_size) {
  stopifnot(
    is.list(chunk),
    length(chunk) > 0,
    total_size > 0
  )
  # ...
}

# R: explicit assertion with context
merge_results <- function(left, right) {
  if (ncol(left) != ncol(right)) {
    stop("Internal error: column count mismatch (", ncol(left), " vs ", ncol(right),
         "). This is a bug — please report it.", call. = FALSE)
  }
  # ...
}

Expected: Internal invariants are asserted so bugs surface immediately at the violation site, not three function calls later with a cryptic error.

On failure: If

stopifnot()
messages are too cryptic, switch to explicit
if/stop
with context.

Step 6: Refactor Anti-Patterns

Identify and fix these common anti-patterns:

Anti-pattern 1: Empty tryCatch (swallowing errors)

# BEFORE: Error silently disappears
result <- tryCatch(
  parse_data(input),
  error = function(e) NULL
)

# AFTER: Log, re-throw, or return a typed error
result <- tryCatch(
  parse_data(input),
  error = function(e) {
    cli::cli_abort("Failed to parse input: {e$message}", parent = e)
  }
)

Anti-pattern 2: Default values masking bad input

# BEFORE: Caller never knows their input was ignored
process <- function(x = 10) {
  if (!is.numeric(x)) x <- 10  # silently replaces bad input
  x * 2
}

# AFTER: Tell the caller about the problem
process <- function(x = 10) {
  if (!is.numeric(x)) {
    stop("'x' must be numeric, got ", class(x)[[1]], call. = FALSE)
  }
  x * 2
}

Anti-pattern 3: suppressWarnings as a fix

# BEFORE: Hiding the symptom instead of fixing the cause
result <- suppressWarnings(as.numeric(user_input))

# AFTER: Validate explicitly, handle the expected case
if (!grepl("^-?\\d+\\.?\\d*$", user_input)) {
  stop("Expected a number, got: '", user_input, "'", call. = FALSE)
}
result <- as.numeric(user_input)

Anti-pattern 4: Catch-all exception handlers

# BEFORE: Every error treated the same
tryCatch(
  complex_operation(),
  error = function(e) message("Something went wrong")
)

# AFTER: Handle specific conditions, let unexpected ones propagate
tryCatch(
  complex_operation(),
  custom_validation_error = function(e) {
    cli::cli_warn("Validation issue: {e$message}")
    fallback_value
  }
  # Unexpected errors propagate naturally
)

Expected: Anti-patterns are replaced with explicit validation or specific error handling.

On failure: If removing a

tryCatch
causes cascading failures, the upstream code has a validation gap. Fix the source, not the symptom.

Step 7: Validate the Fail-Early Refactoring

Run the test suite to confirm error paths work correctly:

# Verify error messages are triggered
testthat::expect_error(calculate_summary("not_a_df"), "must be a data frame")
testthat::expect_error(calculate_summary(data.frame()), "at least one row")
testthat::expect_error(calculate_summary(mtcars, trim_pct = 2), "between 0 and 0.5")

# Verify valid inputs still work
testthat::expect_no_error(calculate_summary(mtcars, method = "mean"))
# Run full test suite
Rscript -e "devtools::test()"

Expected: All tests pass. Error-path tests confirm that bad input triggers the expected error message.

On failure: If existing tests relied on silent failures (e.g., returning NULL on bad input), update them to expect the new error.

Validation

  • Every public function validates its inputs before doing work
  • Error messages answer: what failed, where, why, and how to fix
  • stop()
    is used for conditions that produce incorrect results
  • warning()
    is used only for degraded-but-valid outcomes
  • No empty
    tryCatch
    blocks that swallow errors silently
  • No
    suppressWarnings()
    used as a substitute for proper validation
  • No default values that silently mask invalid input
  • Internal invariants use
    stopifnot()
    or explicit assertions
  • Error-path tests exist for each validation guard
  • Test suite passes after refactoring

Common Pitfalls

  • Validating too deep: Validate at trust boundaries (public API), not in every internal helper. Over-validation adds noise and hurts performance.
  • Error messages without context:
    "Invalid input"
    forces the caller to guess. Always include the parameter name, the expected type/range, and the actual value received.
  • Using warning() when you mean stop(): If the function returns garbage after the warning, the caller gets a wrong answer silently. Use
    stop()
    and let the caller decide how to handle it.
  • Swallowing errors in tryCatch:
    tryCatch(..., error = function(e) NULL)
    hides bugs. If you must catch, log or re-throw with added context.
  • Forgetting call. = FALSE: In R,
    stop("msg")
    includes the call by default, which is noisy for end users. Use
    call. = FALSE
    in user-facing functions.
    cli::cli_abort()
    does this automatically.
  • Validating in tests instead of code: Tests verify behavior but do not protect production callers. Validation belongs in the function itself.
  • Wrong R binary on hybrid systems: On WSL or Docker,
    Rscript
    may resolve to a cross-platform wrapper instead of native R. Check with
    which Rscript && Rscript --version
    . Prefer the native R binary (e.g.,
    /usr/local/bin/Rscript
    on Linux/WSL) for reliability. See Setting Up Your Environment for R path configuration.

Related Skills

  • write-testthat-tests
    - write tests that verify error paths
  • review-pull-request
    - review code for missing validation and silent failures
  • review-software-architecture
    - assess error-handling strategy at the system level
  • create-skill
    - create new skills following the agentskills.io standard
  • security-audit-codebase
    - security-focused review that overlaps with input validation