regex-puzzles-skill

name	regex-puzzles-skill
description	Construct correct regular expressions for common text matching and validation challenges.
version	0.1.0
source_model	mock-strong
extracted_from	{"total_traces":10,"passed_traces":10,"failed_traces":0,"extractor":"manual@0.1","extracted_at":"2026-01-01T00:00:00Z"}
triggers	["regex","pattern","match","validate"]
domains	["regex-puzzles"]
license	Apache-2.0

When to use

Apply this skill when you need to construct or debug a regular expression for text matching, validation, or extraction. Covers: IPv4 addresses, email validation, phone extraction, password strength lookaheads, non-greedy quantifiers, named groups, backreferences, negated character classes, anchors, and Unicode properties.

Procedure

1. Identify what needs to be matched — literal text, patterns, or structure
2. Choose anchoring strategy — full-string match (^...$) vs substring search
3. Build the pattern incrementally from left to right
4. Use character classes for sets of allowed characters
5. Apply quantifiers (*, +, {n,m}) with greedy/non-greedy as needed
6. Use groups for capturing and backreferences
7. Add lookaheads/lookbehinds for zero-width assertions
8. Test against both matching and non-matching inputs
9. Consider edge cases: empty strings, Unicode, special characters

Pattern recipes

• IPv4: Match each octet as (25[0-5]|2[0-4]\d|[01]?\d\d?) separated by \.
• Email: Use [a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}
• Phone: Account for optional parens and separators: \(?\d{3}\)?[-.\s]?\d{3}[-.\s]?\d{4}
• Password lookahead: Chain (?=.*[A-Z])(?=.*\d)(?=.*[!@#$%]) before .{8,}
• Non-greedy: Use .*? instead of .* to match shortest possible
• Named groups: Use (?P<name>...) syntax for readable captures
• Backreference: Capture with (\w+) then match repeat with \1
• Negation: Use [^...] to exclude characters from a class
• Anchors: Use ^ and $ (with multiline flag if needed)
• Unicode: Use \p{L} for any Unicode letter

Examples

match-ipv4-address

Pattern: ^((25[0-5]|2[0-4]\d|[01]?\d\d?)\.){3}(25[0-5]|2[0-4]\d|[01]?\d\d?)$

password-lookahead

Pattern: ^(?=.*[A-Z])(?=.*\d)(?=.*[!@#$%^&*]).{8,}$

backreference-repeated-word

Pattern: \b(\w+)\s+\1\b

Anti-patterns

• Do not use .* when you mean .*? — greedy matching causes over-capture
• Do not forget to escape literal dots — . matches any character without escaping
• Do not nest quantifiers without bounds (e.g., (a+)+) — causes catastrophic backtracking
• Do not use regex for full HTML/XML parsing — use a proper parser instead
• Do not assume ASCII-only input — use Unicode-aware patterns when appropriate