Features
Bundled prompts
8 optimized translation prompts for Hadith, Fiqh, Tafsir, and more with strongly-typed access
Translation validation
Catch LLM hallucinations like malformed segment IDs, Arabic leaks, and forbidden terms
Prompt stacking
Master and specialized prompts combined automatically for optimal results
Gold standards
High-fidelity reference dataset for benchmarking translation quality
Bundled prompts
The library includes 8 specialized prompts optimized through rigorous testing:| Prompt ID | Name | Use case |
|---|---|---|
master_prompt | Master Prompt | Universal grounding rules |
hadith | Hadith | Isnad-heavy texts, Sharh |
fiqh | Fiqh | Legal terminology |
tafsir | Tafsir | Quranic exegesis |
fatawa | Fatawa | Q&A format |
encyclopedia_mixed | Encyclopedia Mixed | Polymath works |
jarh_wa_tadil | Jarh Wa Tadil | Narrator criticism |
usul_al_fiqh | Usul Al Fiqh | Legal methodology |
Translation validation
The validation system catches common LLM translation failures:- Invalid marker format: Malformed segment IDs like
B1234$5orP123ab - Arabic leak: Arabic script in output (except ﷺ)
- Invented IDs: Segment IDs that don’t exist in the source corpus
- Missing ID gaps: Skipped segments between consecutive IDs
- Duplicate IDs: Same segment ID appearing multiple times
- Truncated segments: Incomplete translations marked with
...or[INCOMPLETE] - Collapsed speakers: Speaker labels appearing mid-line instead of on new lines
- Empty parentheses: Excessive
()patterns indicating failed term-pairs
Prompt stacking
Prompt stacking combines the universal master prompt with specialized domain addons:- ALA-LC transliteration with diacritics
- Plain text output (no Markdown)
- Segment ID integrity (no invented or skipped IDs)
- Domain-specific conventions (e.g., narrator chains for Hadith)
Next steps
Installation
Install wobble-bibble with npm, yarn, pnpm, or bun
Quickstart
Get started with prompts and validation in 5 minutes