Wordle 5 helper is a 5-letter word solver that ingests your attempts and tile feedback to control a candidate-word filter and a guess-ranking engine. It combines a Wordle solver workflow with a 5-letter word finder, using constraint propagation across positions, included/excluded letters, and repeat-letter rules. The system can run as a Wordle assistant for both hinting and full auto-guess mode, producing a short list of remaining answers plus a prioritized next-guess set based on letter frequency and coverage.

• Accepts per-try input as WORD + pattern (e.g., CRANE + GYBBY)
• Supports green/yellow/gray states and mixed results per letter
• Handles per-position locking (green) and disallow lists (yellow/gray)
• Tracks “must include” letters and “must not include” letters
• Validates input length and enforces 5-letter normalization

Feature intent: Convert human gameplay feedback into machine constraints that can be applied deterministically to a dictionary of 5-letter words.

• Applies positional matches for green tiles
• Blocks letters in specific positions for yellow tiles
• Removes candidates containing banned letters (gray with repeat-letter logic)
• Enforces minimum counts for letters confirmed present
• Supports strict mode (only legal candidates) vs relaxed mode (exploration guesses)

In-game behavior: Each new row of feedback shrinks the answer space; the output updates instantly to reflect only words consistent with all known constraints.

• Distinguishes “absent” vs “overused” letters across a guess
• Implements min/max count bounds when a letter appears multiple times
• Avoids false eliminations caused by partial grays with repeats
• Prefers candidates matching established multiplicity patterns
• Flags ambiguous cases where feedback implies count ranges

Feature intent: Keep filtering accurate when Wordle feedback involves repeated letters, which is where simple solvers commonly break.

• Scores candidates by letter frequency across remaining pool
• Boosts guesses that test many unseen letters (coverage optimization)
• Optional position-frequency scoring for tighter endgames
• Penalizes repeated letters when exploring
• Offers two lists: “Best Guess” (coverage) and “Best Answer” (likelihood)

In-game behavior: Early guesses tend to maximize information; late guesses pivot to finishing, prioritizing the most probable solution.

• Hint mode: returns top-N suggestions without committing
• Auto-guess: selects the next guess and advances a simulated attempt stack
• Configurable attempt limit (default 6)
• Can stop on a threshold (e.g., when ≤3 candidates remain)
• Exposes the rationale for each chosen guess (score breakdown)

Feature intent: Support both players who want gentle guidance and users who want the tool to fully solve the board.

• Loads a base 5-letter dictionary and optional “answer-only” list
• Allows custom word lists (regional spellings, themed lists)
• Toggles for allowing proper nouns, archaic words, and variants
• Deduplication and normalization pipeline (case, accents, punctuation)
• Fast indexing for repeated daily use

In-game behavior: Output stays consistent with the word set you consider valid, preventing “solver says a word Wordle won’t accept” scenarios.

• Shows remaining candidate count and a preview slice
• Displays constraint summary (locked letters, excluded letters, must-include set)
• Marks why a candidate is valid (position checks, count checks)
• Highlights which letters a suggested guess will test
• Provides export-friendly outputs (plain list, CSV-like lines)

Feature intent: Make results auditable—users can see exactly what the solver believes and why.

flowchart TD
A[User Attempts] --> B[Tile Feedback Parser]
B --> C[Constraint Store]
C --> D[Candidate Filter]
D --> E[Remaining Candidates]
E --> F[Ranking Engine]
F --> G[Hint Output]
F --> H[Auto-Guess Loop]

How do I enter a guess and feedback? Use a 5-letter guess plus a 5-char pattern (G=green, Y=yellow, B=gray), e.g., CRANE GYBBB.

Does it handle repeated letters correctly? Yes. It tracks minimum/maximum counts inferred from mixed feedback so candidates aren’t removed incorrectly.

What’s the difference between “Best Guess” and “Best Answer”? “Best Guess” optimizes information gain; “Best Answer” ranks the most likely solutions among remaining candidates.

Can it use only official answer words? Yes. You can switch to an answer-only list to avoid suggestions that are accepted guesses but never solutions.

Will it still suggest exploration words not in the remaining candidates? In coverage mode, it can propose probe guesses (optional) to test more letters, even if they can’t be the answer.

Can I run it in strict mode? Strict mode outputs only valid candidates as next guesses; relaxed mode allows probe guesses for faster narrowing.

• Attempt + tile feedback parser (green/yellow/gray)
• Hard-rule candidate filtering with constraint store
• Duplicate-letter count bounds and repeat handling
• Dual ranking: information coverage vs solution likelihood
• Hint mode and auto-guess solving loop
• Configurable dictionaries (answer-only, custom lists)
• Explainable output with constraint and scoring visibility