completeness-edge-cases — Edge Case Coverage

Severity: INFO · Auto-fix: No · Category: ✅ Completeness

What It Does

Fires when a prompt contains a task verb but provides no guidance on error or edge-case handling. Prompts that only describe the happy path produce brittle behavior on unexpected inputs.

Trigger Conditions

All three must be true:

1. Task verb present — the prompt contains write, create, implement, build, or generate
2. No edge-case keywords — none of: edge case, error, exception, invalid, empty, null, missing
3. Prompt is more than 100 characters

One finding is emitted for the whole prompt (line -).

Keyword-based detection

The rule matches any occurrence of the keyword anywhere in the prompt text. "Return null on empty input" contains null and empty, so it suppresses the finding. You don't need a dedicated <edge_cases> section — as long as one of the suppression keywords appears, the rule is satisfied.

Examples

No edge-case coverage

Write a function that parses a date string and returns an ISO 8601 date.
The input will be in MM/DD/YYYY format. Return the formatted date.

Finding:

[ INFO ] completeness-edge-cases (line -)
  Consider specifying how to handle edge cases and errors.

With edge-case coverage — inline keyword suppresses the rule

Write a function that parses a date string and returns an ISO 8601 date.
The input will be in MM/DD/YYYY format.
Return null if the input is empty or invalid.
Raise a ValueError if the format cannot be parsed.

(Contains null, empty, invalid, error → rule does not fire)

With explicit edge-case section

Write a function that parses a date string and returns an ISO 8601 date.
The input will be in MM/DD/YYYY format.

Edge cases:
- Empty string → return None
- Invalid format → raise ValueError with the original input
- Two-digit year → treat as 2000s if ≤ 25, else 1900s
- Null input → raise TypeError

(Contains edge case, empty, invalid, null → rule does not fire)

Common Edge Cases to Address

For parsing tasks:

• Empty or null input
• Malformed or unexpected format
• Boundary values (min/max length, date ranges, number limits)
• Encoding issues (non-ASCII, BOM characters)

For transformation tasks:

• Input with missing required fields
• Input with extra/unknown fields
• Type mismatches

For generation tasks:

• What to do when there is not enough context
• Fallback format if the preferred format cannot be produced
• Maximum output length behavior

For retrieval/lookup tasks:

• Not found — return null, empty list, or raise an error?
• Multiple matches — return all, return first, or raise?
• Partial match — return closest or nothing?

False Positives

Short prompts — the 100-character threshold prevents the rule from firing on terse one-liners like "Write a haiku about autumn." which genuinely don't need edge-case guidance.

Creative/open-ended prompts — "Create a story about a detective" doesn't benefit from null/empty handling. Disable the rule for creative writing prompts:

rules:
  completeness_edge_cases: false

The keyword appears in a different context — "Generate an error message for the user" contains error, which suppresses the rule even though no edge cases are defined. This is a known limitation of keyword-based detection.

Configuration

rules:
  completeness_edge_cases: true   # default: true

Disable:

rules:
  completeness_edge_cases: false

The severity is fixed at INFO and cannot be promoted via config — use --fail-level info to make it block CI.