Skip to main content
Wobble-bibble provides 8 specialized translation prompts for different types of Islamic texts. Each prompt combines a master prompt with domain-specific rules to guide LLM translation.

​
Getting Available Prompts

Use getPrompts() to retrieve all available prompts at once: 
import { getPrompts } from 'wobble-bibble';

const prompts = getPrompts();

prompts.forEach(prompt => {
  console.log(`${prompt.id}: ${prompt.name}`);
  console.log(`Is Master: ${prompt.isMaster}`);
  console.log(`Content length: ${prompt.content.length} chars`);
});
All addon prompts (like hadith, fiqh, tafsir) are automatically stacked with the master prompt. You don’t need to combine them manually.

​
Getting a Specific Prompt

For most use cases, you’ll want to retrieve a specific prompt by ID: 
import { getPrompt } from 'wobble-bibble';

// Get the Hadith prompt (master + hadith addon)
const hadithPrompt = getPrompt('hadith');

console.log(hadithPrompt.id);        // 'hadith'
console.log(hadithPrompt.name);      // 'Hadith'
console.log(hadithPrompt.isMaster);  // false
console.log(hadithPrompt.content);   // Full stacked prompt text

​
Type Safety

The getPrompt() function is strongly typed, so TypeScript will catch invalid prompt IDs at compile time: 
import type { PromptId } from 'wobble-bibble';

// ✅ Valid - TypeScript knows these IDs exist
const valid: PromptId[] = [
  'master_prompt',
  'hadith',
  'fiqh',
  'tafsir',
  'fatawa',
  'encyclopedia_mixed',
  'jarh_wa_tadil',
  'usul_al_fiqh'
];

// ❌ Invalid - TypeScript error
const invalid = getPrompt('unknown_prompt');

​
Available Prompt IDs

Use getPromptIds() to get a list of all available prompt IDs (useful for dropdowns or validation): 
import { getPromptIds } from 'wobble-bibble';

const ids = getPromptIds();
console.log(ids); // ['master_prompt', 'hadith', 'fiqh', ...]

​
Using Prompts with LLMs

Here’s a complete example of using a prompt with an LLM to translate Arabic segments: 
import { getPrompt, formatExcerptsForPrompt } from 'wobble-bibble';
import OpenAI from 'openai';

const openai = new OpenAI();

const segments = [
  { id: 'P1234', text: 'نعم، هذا صحيح.' },
  { id: 'P1235', text: 'قال الشيخ: لا بأس.' }
];

// Get the Hadith prompt
const prompt = getPrompt('hadith');

// Format segments for the LLM
const userMessage = formatExcerptsForPrompt(segments, '');

const completion = await openai.chat.completions.create({
  model: 'gpt-4',
  messages: [
    { role: 'system', content: prompt.content },
    { role: 'user', content: userMessage }
  ]
});

const translation = completion.choices[0].message.content;

​
Getting Just the Content

If you only need the prompt text (not the metadata), use getStackedPrompt(): 
import { getStackedPrompt } from 'wobble-bibble';

const promptText = getStackedPrompt('fiqh');
// Returns: string with full prompt content

​
Prompt Structure

1

Master Prompt

The master prompt contains universal rules that apply to all translations:
  • Output format requirements (plain text, no Markdown)
  • ID integrity rules (preserve segment IDs exactly)
  • Transliteration standards (ALA-LC with diacritics)
  • Forbidden patterns (no Arabic script except ﷺ)
2

Addon Prompt

Each specialized addon adds domain-specific rules:
  • Hadith: Isnad chain formatting, narrator name handling
  • Fiqh: Legal terminology, ruling formats
  • Tafsir: Quranic verse references, exegesis style
  • Fatawa: Q&A structure, speaker label preservation
3

Automatic Stacking

When you call getPrompt('hadith'), wobble-bibble automatically:
  1. Loads the master prompt
  2. Loads the hadith addon
  3. Combines them with a newline: master + '\n' + addon
  4. Returns the stacked result

​
Choosing the Right Prompt

Use for texts containing:
  • Isnad chains (e.g., “عن أبي هريرة عن النبي”)
  • Hadith commentary (Sharh)
  • Narrator criticism embedded in hadith discussions
const prompt = getPrompt('hadith');
Use for texts containing:
  • Verse-by-verse Quran commentary
  • Quranic references and citations
  • Linguistic analysis of Quranic words
const prompt = getPrompt('tafsir');
Use for texts containing:
  • Question and answer dialogues
  • Speaker labels (السائل, الشيخ)
  • Multi-turn conversations
const prompt = getPrompt('fatawa');
Use for texts that mix multiple genres:
  • Works by scholars like Ibn Taymiyyah or Al-Albani
  • Texts that switch between hadith, fiqh, and other topics
  • No mode tags will appear in output
const prompt = getPrompt('encyclopedia_mixed');

​
Working with the Master Prompt

If you need to create a custom addon, you can access the master prompt directly: 
import { getMasterPrompt, stackPrompts } from 'wobble-bibble';

const masterPrompt = getMasterPrompt();
const customAddon = `
CUSTOM RULE:
Always translate technical terms with parenthetical glosses.
`;

const customPrompt = stackPrompts(masterPrompt, customAddon);
The master prompt should never be modified. It contains critical rules for ID integrity, output formatting, and transliteration standards. Only create custom addons that extend these rules.

​
Practical Examples

​
Building a Translation UI

import { getPromptIds, getPrompt } from 'wobble-bibble';

function PromptSelector() {
  const [selectedId, setSelectedId] = useState<PromptId>('hadith');
  const ids = getPromptIds();
  
  const handleTranslate = async () => {
    const prompt = getPrompt(selectedId);
    // Use prompt.content with your LLM...
  };
  
  return (
    <select value={selectedId} onChange={e => setSelectedId(e.target.value)}>
      {ids.map(id => (
        <option key={id} value={id}>{id}</option>
      ))}
    </select>
  );
}

​
Batch Translation with Different Prompts

import { getPrompt, formatExcerptsForPrompt } from 'wobble-bibble';

interface TranslationJob {
  promptType: PromptId;
  segments: Segment[];
}

async function processBatch(jobs: TranslationJob[]) {
  const results = [];
  
  for (const job of jobs) {
    const prompt = getPrompt(job.promptType);
    const userMessage = formatExcerptsForPrompt(job.segments, '');
    
    // Call your LLM here with prompt.content and userMessage
    const translation = await callLLM(prompt.content, userMessage);
    
    results.push({
      promptType: job.promptType,
      translation
    });
  }
  
  return results;
}

​
Next Steps

Validating Output

Learn how to validate LLM translations for errors

Custom Prompts

Create custom prompts for specialized use cases

Build docs developers (and LLMs) love

Get started for freeTalk to us