diff --git a/data/urdu_islamic_terms.json b/data/urdu_islamic_terms.json
new file mode 100644
index 0000000..4256992
--- /dev/null
+++ b/data/urdu_islamic_terms.json
@@ -0,0 +1,227 @@
+{
+  "version": "1.0.0",
+  "language": "ur",
+  "description": "Curated core Islamic terminology for Urdu query processing.",
+  "terms": [
+    {
+      "id": "ur-ibadat-salah",
+      "urdu_term": "نماز",
+      "arabic_original": "صلاة",
+      "transliteration": "namaz / salah",
+      "english_equivalent": "ritual prayer",
+      "category": "ibadat",
+      "definition_ur": "روزانہ مقررہ اوقات میں ادا کی جانے والی فرض عبادت۔",
+      "variants": ["صلوٰۃ", "صلاۃ", "صلاح", "salah", "salat", "namaz"]
+    },
+    {
+      "id": "ur-ibadat-wudu",
+      "urdu_term": "وضو",
+      "arabic_original": "وضوء",
+      "transliteration": "wudu",
+      "english_equivalent": "ritual ablution",
+      "category": "ibadat",
+      "definition_ur": "نماز وغیرہ کے لیے مقررہ طریقے سے حاصل کی جانے والی طہارت۔",
+      "variants": ["وضوء", "وضو کرنا", "wudu", "wudhu"]
+    },
+    {
+      "id": "ur-ibadat-fasting",
+      "urdu_term": "روزہ",
+      "arabic_original": "صوم",
+      "transliteration": "roza / sawm",
+      "english_equivalent": "fasting",
+      "category": "ibadat",
+      "definition_ur": "نیت کے ساتھ فجر سے غروب آفتاب تک مفطرات سے رکنا۔",
+      "variants": ["صوم", "صیام", "roza", "sawm", "siyam"]
+    },
+    {
+      "id": "ur-ibadat-zakat",
+      "urdu_term": "زکوٰۃ",
+      "arabic_original": "زكاة",
+      "transliteration": "zakat",
+      "english_equivalent": "obligatory almsgiving",
+      "category": "ibadat",
+      "definition_ur": "شرائط پوری ہونے پر مال کا مقررہ حصہ مستحقین کو دینا۔",
+      "variants": ["زکات", "زکاۃ", "زكاة", "zakat", "zakah"]
+    },
+    {
+      "id": "ur-ibadat-hajj",
+      "urdu_term": "حج",
+      "arabic_original": "حج",
+      "transliteration": "hajj",
+      "english_equivalent": "pilgrimage to Makkah",
+      "category": "ibadat",
+      "definition_ur": "استطاعت رکھنے والے مسلمان کا مقررہ ایام میں بیت اللہ کا شرعی حج۔",
+      "variants": ["حج بیت اللہ", "hajj"]
+    },
+    {
+      "id": "ur-quran-quran",
+      "urdu_term": "قرآن مجید",
+      "arabic_original": "القرآن الكريم",
+      "transliteration": "Quran Majid",
+      "english_equivalent": "the Noble Quran",
+      "category": "quran",
+      "definition_ur": "اللہ تعالیٰ کی آخری نازل کردہ کتاب۔",
+      "variants": ["قرآن", "قران", "قرآن کریم", "القرآن", "Quran", "Qur'an"]
+    },
+    {
+      "id": "ur-quran-tafsir",
+      "urdu_term": "تفسیر",
+      "arabic_original": "تفسير",
+      "transliteration": "tafsir",
+      "english_equivalent": "Quranic exegesis",
+      "category": "quran",
+      "definition_ur": "قرآن مجید کے معانی اور مطالب کی علمی وضاحت۔",
+      "variants": ["تفسير", "tafsir"]
+    },
+    {
+      "id": "ur-hadith-hadith",
+      "urdu_term": "حدیث",
+      "arabic_original": "حديث",
+      "transliteration": "hadith",
+      "english_equivalent": "Prophetic narration",
+      "category": "hadith",
+      "definition_ur": "رسول اللہ ﷺ کے اقوال، افعال، تقریرات یا اوصاف کی روایت۔",
+      "variants": ["حديث", "احادیث", "hadith", "hadees"]
+    },
+    {
+      "id": "ur-hadith-sunnah",
+      "urdu_term": "سنت",
+      "arabic_original": "سنة",
+      "transliteration": "sunnah",
+      "english_equivalent": "Prophetic practice",
+      "category": "hadith",
+      "definition_ur": "رسول اللہ ﷺ کا منقول طریقہ اور تعلیم۔",
+      "variants": ["سنۃ", "سنة", "sunnah", "sunna"]
+    },
+    {
+      "id": "ur-fiqh-fiqh",
+      "urdu_term": "فقہ",
+      "arabic_original": "فقه",
+      "transliteration": "fiqh",
+      "english_equivalent": "Islamic jurisprudence",
+      "category": "fiqh",
+      "definition_ur": "تفصیلی دلائل سے عملی شرعی احکام جاننے کا علم۔",
+      "variants": ["فقه", "fiqh"]
+    },
+    {
+      "id": "ur-fiqh-fatwa",
+      "urdu_term": "فتویٰ",
+      "arabic_original": "فتوى",
+      "transliteration": "fatwa",
+      "english_equivalent": "formal legal opinion",
+      "category": "fiqh",
+      "definition_ur": "کسی شرعی سوال پر اہل علم کی مدلل رائے۔",
+      "variants": ["فتوی", "فتوى", "فتاویٰ", "fatwa"]
+    },
+    {
+      "id": "ur-fiqh-shariah",
+      "urdu_term": "شریعت",
+      "arabic_original": "شريعة",
+      "transliteration": "shariah",
+      "english_equivalent": "Islamic revealed law and way",
+      "category": "fiqh",
+      "definition_ur": "اسلام کا الٰہی ہدایت پر مبنی مجموعی طریقۂ زندگی۔",
+      "variants": ["شریعہ", "شريعة", "sharia", "shariah"]
+    },
+    {
+      "id": "ur-fiqh-hanafi",
+      "urdu_term": "فقہ حنفی",
+      "arabic_original": "المذهب الحنفي",
+      "transliteration": "Hanafi fiqh",
+      "english_equivalent": "Hanafi school of jurisprudence",
+      "category": "fiqh",
+      "definition_ur": "امام ابو حنیفہؒ سے منسوب اہل سنت کا فقہی مذہب۔",
+      "variants": ["حنفی", "مذہب حنفی", "Hanafi"]
+    },
+    {
+      "id": "ur-belief-tawhid",
+      "urdu_term": "توحید",
+      "arabic_original": "توحيد",
+      "transliteration": "tawhid",
+      "english_equivalent": "oneness of Allah",
+      "category": "aqidah",
+      "definition_ur": "اللہ تعالیٰ کو ذات، صفات اور عبادت میں یکتا ماننا۔",
+      "variants": ["توحيد", "tawhid", "tauheed"]
+    },
+    {
+      "id": "ur-belief-aqidah",
+      "urdu_term": "عقیدہ",
+      "arabic_original": "عقيدة",
+      "transliteration": "aqidah",
+      "english_equivalent": "creed or belief",
+      "category": "aqidah",
+      "definition_ur": "دین کے بنیادی ایمانی تصورات اور پختہ اعتقادات۔",
+      "variants": ["عقیدۂ", "عقائد", "عقيدة", "aqidah", "aqeedah"]
+    },
+    {
+      "id": "ur-ethics-taqwa",
+      "urdu_term": "تقویٰ",
+      "arabic_original": "تقوى",
+      "transliteration": "taqwa",
+      "english_equivalent": "God-consciousness",
+      "category": "akhlaq",
+      "definition_ur": "اللہ کی جواب دہی کا شعور رکھتے ہوئے گناہ سے بچنا۔",
+      "variants": ["تقوی", "تقوى", "taqwa"]
+    },
+    {
+      "id": "ur-ethics-sabr",
+      "urdu_term": "صبر",
+      "arabic_original": "صبر",
+      "transliteration": "sabr",
+      "english_equivalent": "patience and steadfastness",
+      "category": "akhlaq",
+      "definition_ur": "اطاعت، مصیبت اور گناہ سے اجتناب میں ثابت قدمی۔",
+      "variants": ["sabr"]
+    },
+    {
+      "id": "ur-ethics-dua",
+      "urdu_term": "دعا",
+      "arabic_original": "دعاء",
+      "transliteration": "dua",
+      "english_equivalent": "supplication",
+      "category": "ibadat",
+      "definition_ur": "اللہ تعالیٰ سے حاجت اور مدد طلب کرنا۔",
+      "variants": ["دعاء", "دُعا", "dua", "du'a"]
+    },
+    {
+      "id": "ur-family-nikah",
+      "urdu_term": "نکاح",
+      "arabic_original": "نكاح",
+      "transliteration": "nikah",
+      "english_equivalent": "Islamic marriage contract",
+      "category": "family",
+      "definition_ur": "مرد اور عورت کے درمیان شرعی عقد ازدواج۔",
+      "variants": ["نكاح", "nikah", "nikaah"]
+    },
+    {
+      "id": "ur-family-talaq",
+      "urdu_term": "طلاق",
+      "arabic_original": "طلاق",
+      "transliteration": "talaq",
+      "english_equivalent": "divorce",
+      "category": "family",
+      "definition_ur": "شرعی طریقے سے نکاح ختم کرنے کا عمل، جس کے احکام تفصیلی ہیں۔",
+      "variants": ["talaq", "divorce"]
+    },
+    {
+      "id": "ur-honorific-allah",
+      "urdu_term": "اللہ تعالیٰ",
+      "arabic_original": "الله تعالى",
+      "transliteration": "Allah Ta'ala",
+      "english_equivalent": "Allah, the Exalted",
+      "category": "honorific",
+      "definition_ur": "اللہ کے نام کے ساتھ ادب و تعظیم کا معروف اسلوب۔",
+      "variants": ["اللہ", "الله", "اللہ تعالٰی", "Allah"]
+    },
+    {
+      "id": "ur-honorific-prophet",
+      "urdu_term": "رسول اللہ ﷺ",
+      "arabic_original": "رسول الله صلى الله عليه وسلم",
+      "transliteration": "Rasul Allah, sallallahu alayhi wa sallam",
+      "english_equivalent": "Messenger of Allah, peace and blessings be upon him",
+      "category": "honorific",
+      "definition_ur": "نبی کریم محمد ﷺ کے ذکر کا باادب اسلوب۔",
+      "variants": ["رسول اللہ", "نبی کریم ﷺ", "حضرت محمد ﷺ", "صلی اللہ علیہ وسلم", "PBUH"]
+    }
+  ]
+}
diff --git a/tests/test_urdu_processor.py b/tests/test_urdu_processor.py
new file mode 100644
index 0000000..fdd2445
--- /dev/null
+++ b/tests/test_urdu_processor.py
@@ -0,0 +1,75 @@
+"""Offline tests for Urdu normalization, terminology, and mixed-script processing."""
+
+from fastapi.testclient import TestClient
+from fastapi import FastAPI
+
+from urdu_processor import (
+    analyze_script,
+    extract_islamic_terms,
+    normalize_urdu,
+    process_urdu,
+    router,
+    tokenize_urdu,
+    transliterate_urdu,
+)
+
+
+def test_normalizes_arabic_keyboard_variants_and_nastaliq_spacing() -> None:
+    assert normalize_urdu("  زكاة\u200c  كي  حكم؟ ") == "زکاۃ کی حکم؟"
+
+
+def test_diacritics_are_preserved_by_default_and_optionally_removed() -> None:
+    text = "قُرْآن"
+    assert "ُ" in normalize_urdu(text)
+    assert normalize_urdu(text, preserve_diacritics=False) == "قرآن"
+
+
+def test_multiword_islamic_terms_are_single_tokens() -> None:
+    tokens = tokenize_urdu("فقہ حنفی میں زکوٰۃ کا حکم")
+    assert tokens[0].text == "فقہ حنفی"
+    assert tokens[0].kind == "islamic_term"
+    assert tokens[0].term_id == "ur-fiqh-hanafi"
+
+
+def test_recognizes_urdu_arabic_and_latin_variants_without_duplicates() -> None:
+    terms = extract_islamic_terms("وضو اور wudu کے بعد نماز salah")
+    ids = [term.id for term in terms]
+    assert ids.count("ur-ibadat-wudu") == 1
+    assert ids.count("ur-ibadat-salah") == 1
+
+
+def test_mixed_script_profile() -> None:
+    profile = analyze_script("زکوٰۃ nisab 2.5%")
+    assert profile.mixed_script is True
+    assert profile.dominant_script == "mixed"
+    assert profile.urdu_arabic_characters > 0
+    assert profile.latin_characters > 0
+
+
+def test_curated_transliteration_is_preferred_for_terms() -> None:
+    result = transliterate_urdu("زکوٰۃ اور نماز")
+    assert "zakat" in result
+    assert "namaz / salah" in result
+
+
+def test_full_analysis_supplies_urdu_generation_and_citation_guidance() -> None:
+    result = process_urdu("Quran میں صبر کے متعلق کیا حکم ہے؟")
+    assert result.normalized_text
+    assert result.script_profile.mixed_script is True
+    assert any(term.id == "ur-quran-quran" for term in result.recognized_terms)
+    assert "قرآنی آیات" in result.generation_guidance
+    assert "حوالہ" in result.generation_guidance
+
+
+def test_router_process_and_term_search() -> None:
+    app = FastAPI()
+    app.include_router(router)
+    client = TestClient(app)
+
+    processed = client.post("/urdu/process", json={"text": "نماز اور وضو"})
+    assert processed.status_code == 200
+    assert len(processed.json()["recognized_terms"]) == 2
+
+    searched = client.get("/urdu/terms", params={"query": "tawhid"})
+    assert searched.status_code == 200
+    assert searched.json()[0]["id"] == "ur-belief-tawhid"
diff --git a/urdu_processor.py b/urdu_processor.py
new file mode 100644
index 0000000..f70b6fc
--- /dev/null
+++ b/urdu_processor.py
@@ -0,0 +1,360 @@
+"""Urdu Islamic-language processing utilities.
+
+The processor is deterministic and offline.  It normalizes Urdu, Arabic, and
+Persian Unicode variants without changing Quranic Arabic by default, recognizes
+curated Islamic terminology, preserves multi-word terms during tokenization,
+and supplies prompt guidance for an existing multilingual generation pipeline.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+import unicodedata
+from functools import lru_cache
+from pathlib import Path
+from typing import Literal
+
+from fastapi import APIRouter
+from pydantic import BaseModel, Field
+
+router = APIRouter(prefix="/urdu", tags=["urdu"])
+
+DATA_PATH = Path(__file__).resolve().parent / "data" / "urdu_islamic_terms.json"
+
+# Presentation forms are compatibility-normalized by NFKC.  These mappings
+# then fold Arabic/Persian keyboard variants to the standard Urdu code points.
+_URDU_CHAR_MAP = str.maketrans(
+    {
+        "ي": "ی",
+        "ى": "ی",
+        "ئ": "ئ",
+        "ك": "ک",
+        "ة": "ۃ",
+        "ۀ": "ہ",
+        "ه": "ہ",
+        "ھ": "ھ",
+        "ؤ": "ؤ",
+        "ـ": "",
+        "\u200b": "",
+        "\u200c": "",
+        "\u200d": "",
+        "\ufeff": "",
+    }
+)
+
+# Arabic combining marks, Quranic annotation signs, and superscript alif.
+_DIACRITICS_RE = re.compile("[\u0610-\u061a\u064b-\u065f\u0670\u06d6-\u06ed]")
+_WHITESPACE_RE = re.compile(r"[\t\r\f\v ]+")
+_URDU_CHAR_RE = re.compile(r"[\u0600-\u06ff\u0750-\u077f\u08a0-\u08ff]")
+_LATIN_CHAR_RE = re.compile(r"[A-Za-z]")
+_DEVANAGARI_CHAR_RE = re.compile(r"[\u0900-\u097f]")
+_TOKEN_RE = re.compile(
+    r"[\u0600-\u06ff\u0750-\u077f\u08a0-\u08ff]+(?:['’][\u0600-\u06ff]+)*"
+    r"|[A-Za-z]+(?:['’-][A-Za-z]+)*|\d+(?:[.,:/-]\d+)*|[^\s]"
+)
+
+# Conservative character transliteration intended for discoverability and
+# mixed-script search, not for replacing a scholarly transliteration standard.
+_TRANSLITERATION_MAP = {
+    "ا": "a",
+    "آ": "aa",
+    "أ": "a",
+    "إ": "i",
+    "ء": "'",
+    "ؤ": "u",
+    "ئ": "y",
+    "ب": "b",
+    "پ": "p",
+    "ت": "t",
+    "ٹ": "t",
+    "ث": "th",
+    "ج": "j",
+    "چ": "ch",
+    "ح": "h",
+    "خ": "kh",
+    "د": "d",
+    "ڈ": "d",
+    "ذ": "dh",
+    "ر": "r",
+    "ڑ": "r",
+    "ز": "z",
+    "ژ": "zh",
+    "س": "s",
+    "ش": "sh",
+    "ص": "s",
+    "ض": "d",
+    "ط": "t",
+    "ظ": "z",
+    "ع": "'",
+    "غ": "gh",
+    "ف": "f",
+    "ق": "q",
+    "ک": "k",
+    "گ": "g",
+    "ل": "l",
+    "م": "m",
+    "ن": "n",
+    "ں": "n",
+    "و": "w",
+    "ہ": "h",
+    "ۃ": "h",
+    "ھ": "h",
+    "ی": "y",
+    "ے": "e",
+}
+
+
+class UrduIslamicTerm(BaseModel):
+    """A curated Islamic term used in Urdu scholarly and everyday writing."""
+
+    id: str
+    urdu_term: str
+    arabic_original: str
+    transliteration: str
+    english_equivalent: str
+    category: str
+    definition_ur: str
+    variants: list[str] = Field(default_factory=list)
+
+
+class UrduToken(BaseModel):
+    """A token with optional terminology metadata."""
+
+    text: str
+    normalized: str
+    kind: Literal["islamic_term", "urdu", "latin", "number", "punctuation"]
+    term_id: str | None = None
+
+
+class ScriptProfile(BaseModel):
+    """Character-level script statistics for code-switched input."""
+
+    urdu_arabic_characters: int
+    latin_characters: int
+    devanagari_characters: int
+    dominant_script: Literal["urdu_arabic", "latin", "devanagari", "mixed", "none"]
+    mixed_script: bool
+
+
+class UrduAnalysis(BaseModel):
+    """Normalized text and linguistic signals for downstream retrieval/generation."""
+
+    original_text: str
+    normalized_text: str
+    tokens: list[UrduToken]
+    recognized_terms: list[UrduIslamicTerm]
+    script_profile: ScriptProfile
+    transliteration: str
+    generation_guidance: str
+
+
+class UrduProcessRequest(BaseModel):
+    text: str = Field(min_length=1)
+    preserve_diacritics: bool = True
+
+
+class UrduTerminology:
+    """Load and search the bundled Urdu Islamic terminology database."""
+
+    def __init__(self, data_file: Path = DATA_PATH) -> None:
+        self.terms: list[UrduIslamicTerm] = []
+        self._lookup: dict[str, UrduIslamicTerm] = {}
+        if data_file.exists():
+            with data_file.open(encoding="utf-8") as source:
+                payload = json.load(source)
+            for raw in payload.get("terms", []):
+                term = UrduIslamicTerm.model_validate(raw)
+                self.terms.append(term)
+                for form in (term.urdu_term, term.arabic_original, term.transliteration, *term.variants):
+                    key = normalize_urdu(form, preserve_diacritics=False).casefold()
+                    if key:
+                        self._lookup[key] = term
+
+    def lookup(self, text: str) -> UrduIslamicTerm | None:
+        key = normalize_urdu(text, preserve_diacritics=False).casefold()
+        return self._lookup.get(key)
+
+    def search(self, query: str, limit: int = 20) -> list[UrduIslamicTerm]:
+        key = normalize_urdu(query, preserve_diacritics=False).casefold()
+        if not key:
+            return self.terms[:limit]
+        exact = self._lookup.get(key)
+        matches = [
+            term
+            for term in self.terms
+            if key in normalize_urdu(
+                " ".join(
+                    [term.urdu_term, term.arabic_original, term.transliteration, term.english_equivalent, *term.variants]
+                ),
+                preserve_diacritics=False,
+            ).casefold()
+        ]
+        if exact is not None:
+            matches = [exact, *(term for term in matches if term.id != exact.id)]
+        return matches[:limit]
+
+    @property
+    def phrases(self) -> list[str]:
+        forms = [form for form in self._lookup if " " in form]
+        return sorted(forms, key=len, reverse=True)
+
+
+@lru_cache(maxsize=1)
+def get_terminology() -> UrduTerminology:
+    return UrduTerminology()
+
+
+def normalize_urdu(text: str, preserve_diacritics: bool = True) -> str:
+    """Normalize presentation forms, keyboard variants, spacing, and punctuation.
+
+    Diacritics are preserved by default because they can be semantically
+    important in Arabic quotations.  Callers may remove them for retrieval keys.
+    """
+    normalized = unicodedata.normalize("NFKC", text).translate(_URDU_CHAR_MAP)
+    if not preserve_diacritics:
+        normalized = _DIACRITICS_RE.sub("", normalized)
+    normalized = normalized.replace("?", "؟").replace(",", "،")
+    lines = [_WHITESPACE_RE.sub(" ", line).strip() for line in normalized.split("\n")]
+    return "\n".join(line for line in lines if line).strip()
+
+
+def analyze_script(text: str) -> ScriptProfile:
+    urdu_count = len(_URDU_CHAR_RE.findall(text))
+    latin_count = len(_LATIN_CHAR_RE.findall(text))
+    devanagari_count = len(_DEVANAGARI_CHAR_RE.findall(text))
+    populated = sum(count > 0 for count in (urdu_count, latin_count, devanagari_count))
+    if populated > 1:
+        dominant = "mixed"
+    elif urdu_count:
+        dominant = "urdu_arabic"
+    elif latin_count:
+        dominant = "latin"
+    elif devanagari_count:
+        dominant = "devanagari"
+    else:
+        dominant = "none"
+    return ScriptProfile(
+        urdu_arabic_characters=urdu_count,
+        latin_characters=latin_count,
+        devanagari_characters=devanagari_count,
+        dominant_script=dominant,
+        mixed_script=populated > 1,
+    )
+
+
+def _protect_phrases(text: str, terminology: UrduTerminology) -> tuple[str, dict[str, str]]:
+    protected = text
+    replacements: dict[str, str] = {}
+    for index, phrase in enumerate(terminology.phrases):
+        pattern = re.compile(r"(?<![\w\u0600-\u06ff])" + re.escape(phrase) + r"(?![\w\u0600-\u06ff])", re.IGNORECASE)
+        if pattern.search(protected):
+            marker = f"ZZURDUTERM{index}ZZ"
+            protected = pattern.sub(marker, protected)
+            replacements[marker] = phrase
+    return protected, replacements
+
+
+def tokenize_urdu(text: str) -> list[UrduToken]:
+    """Tokenize mixed Urdu text while keeping known multi-word terms intact."""
+    terminology = get_terminology()
+    normalized_text = normalize_urdu(text)
+    protected, replacements = _protect_phrases(normalized_text, terminology)
+    tokens: list[UrduToken] = []
+    for raw in _TOKEN_RE.findall(protected):
+        token_text = replacements.get(raw, raw)
+        normalized = normalize_urdu(token_text, preserve_diacritics=False)
+        term = terminology.lookup(normalized)
+        if term is not None:
+            kind: Literal["islamic_term", "urdu", "latin", "number", "punctuation"] = "islamic_term"
+        elif normalized.replace(".", "").replace(",", "").isdigit():
+            kind = "number"
+        elif _URDU_CHAR_RE.search(normalized):
+            kind = "urdu"
+        elif _LATIN_CHAR_RE.search(normalized):
+            kind = "latin"
+        else:
+            kind = "punctuation"
+        tokens.append(
+            UrduToken(
+                text=token_text,
+                normalized=normalized,
+                kind=kind,
+                term_id=term.id if term else None,
+            )
+        )
+    return tokens
+
+
+def transliterate_urdu(text: str) -> str:
+    """Return a stable Latin transliteration, preferring curated term forms."""
+    terminology = get_terminology()
+    tokens = tokenize_urdu(text)
+    rendered: list[str] = []
+    for token in tokens:
+        term = terminology.lookup(token.normalized)
+        if term is not None:
+            rendered.append(term.transliteration)
+            continue
+        value = "".join(_TRANSLITERATION_MAP.get(char, char) for char in token.normalized)
+        rendered.append(value)
+
+    result = " ".join(rendered)
+    result = re.sub(r"\s+([،۔؟!,:;])", r"\1", result)
+    result = result.replace("،", ",").replace("۔", ".").replace("؟", "?")
+    return re.sub(r"\s+", " ", result).strip()
+
+
+def extract_islamic_terms(text: str) -> list[UrduIslamicTerm]:
+    """Recognize terms once, in textual order, across Urdu and Latin forms."""
+    terminology = get_terminology()
+    found: dict[str, UrduIslamicTerm] = {}
+    for token in tokenize_urdu(text):
+        term = terminology.lookup(token.normalized)
+        if term is not None:
+            found.setdefault(term.id, term)
+    return list(found.values())
+
+
+def build_generation_guidance(terms: list[UrduIslamicTerm], profile: ScriptProfile) -> str:
+    """Create concise instructions for the existing multilingual model pipeline."""
+    guidance = [
+        "جواب واضح، باادب اور فطری اردو میں دیں۔",
+        "قرآنی آیات اور احادیث کے اصل عربی متن کو تبدیل نہ کریں اور حوالہ واضح لکھیں۔",
+        "فقہی اختلاف میں کسی ایک رائے کو بلا وضاحت قطعی یا واحد رائے نہ کہیں۔",
+    ]
+    if terms:
+        labels = "، ".join(f"{term.urdu_term} ({term.transliteration})" for term in terms)
+        guidance.append(f"سوال میں شناخت شدہ اسلامی اصطلاحات: {labels}۔ انہی مستند املا اور معانی کو ملحوظ رکھیں۔")
+    if profile.mixed_script:
+        guidance.append("مخلوط رسم الخط کو سمجھیں، مگر جواب کا بنیادی رسم الخط اردو رکھیں؛ ضروری اصل عربی محفوظ رکھیں۔")
+    return " ".join(guidance)
+
+
+def process_urdu(text: str, preserve_diacritics: bool = True) -> UrduAnalysis:
+    """Run the complete Urdu preprocessing and query-understanding pipeline."""
+    normalized = normalize_urdu(text, preserve_diacritics=preserve_diacritics)
+    profile = analyze_script(normalized)
+    terms = extract_islamic_terms(normalized)
+    return UrduAnalysis(
+        original_text=text,
+        normalized_text=normalized,
+        tokens=tokenize_urdu(normalized),
+        recognized_terms=terms,
+        script_profile=profile,
+        transliteration=transliterate_urdu(normalized),
+        generation_guidance=build_generation_guidance(terms, profile),
+    )
+
+
+@router.post("/process", response_model=UrduAnalysis)
+def process_endpoint(request: UrduProcessRequest) -> UrduAnalysis:
+    """Normalize and analyze an Urdu or Urdu-code-switched query."""
+    return process_urdu(request.text, preserve_diacritics=request.preserve_diacritics)
+
+
+@router.get("/terms", response_model=list[UrduIslamicTerm])
+def search_terms(query: str = "", limit: int = 20) -> list[UrduIslamicTerm]:
+    """Search curated terminology by Urdu, Arabic, transliteration, or English."""
+    safe_limit = min(max(limit, 1), 100)
+    return get_terminology().search(query, safe_limit)