ProGuard Obfuscation Dictionaries

Dictionary	Category	Chars	3-char Combos
Homoglyph ↗ Cross-script visual confusables	Visual	255	16,581,375	Download
Greek ↗ Greek alphabet letters & extended forms	Script	122	1,815,848	Download
Cyrillic ↗ Cyrillic basic & extended letters	Script	296	25,934,336	Download
Latin Extended ↗ Diacritical marks, IPA, Latin Extended	Script	494	120,553,784	Download
Hebrew + Arabic ↗ Hebrew & Arabic with presentation forms	Script	423	75,686,967	Download
East Asian ↗ CJK radicals, Katakana, Hiragana, halfwidth	Script	246	14,886,936	Download
Modifier & Tone ↗ Modifier letters, tone marks, super/subscripts	Visual	244	14,526,784	Download
Indigenous Scripts ↗ Cherokee, Georgian, Armenian, Thai, Lao, etc.	Script	1,146	1,505,060,136	Download
Encoding Chaos ↗ Different display in GBK/Big5/Shift-JIS/EUC-KR	Encoding	575	190,109,375	Download
Minimal Confusable ↗ One best replacement per ASCII letter	Minimal	52	140,608	Download
Ultimate ↗ All scripts merged — maximum character set	Combined	3,056	28,540,399,616	Download

---

How to Use

Download a dictionary file and place it in your Android project (e.g., next to proguard-rules.pro), then add the directive to your ProGuard configuration:

Or use the ADB Pro plugin's R8 Assistant → Dictionary tab to download and apply dictionaries directly from your IDE.

---

About These Dictionaries

Each dictionary contains Unicode characters that are valid Java identifier starts (passing Character.isJavaIdentifierStart()). They exploit visual similarities between characters from different writing systems (homoglyphs) or characters that display differently depending on the system encoding.

For example, the Homoglyph dictionary includes Greek and Cyrillic letters that look identical to Latin letters (e.g., Greek alpha α vs Latin a). The Encoding Chaos dictionary uses characters whose UTF-8 bytes decode to completely different characters when interpreted in GBK, Big5, or Shift-JIS.

All 11 dictionaries are open and free to use, even without the ADB Pro plugin.

Recommendations

• For maximum confusion: Use the Ultimate dictionary (all 3,056 characters merged)
• For a balanced approach: Use Homoglyph + Encoding Chaos together
• For lightweight obfuscation: Use Minimal Confusable (52 characters, one per ASCII letter)
• For encoding-specific confusion: Use Encoding Chaos (characters display differently in CJK environments)