@@ -1,5 +1,6 @@
 # AI service logic with LLM integration (OpenAI/Gemini)
 # Implements 'generate_book_note' and 'get_ai_recommendations'. All recommendations MUST be AI-based.
+
 # Enhanced with comprehensive caching for expensive operations
 
 
@@ -219,6 +220,25 @@ def get_category_books_prompt(category: str, vibe_description: str, count: int =
 - Output the JSON array only.
 """
 
+    @staticmethod
+    def get_vibe_check_prompt(vibe_prompt: str, count: int = 3) -> str:
+        """
+        Prompt for generating hyper-personalized vibe-based book recommendations.
+        """
+        return f"""You are Elara, an emotionally intuitive bookseller.
+A reader is looking for books matching this exact hyper-specific vibe or mood: "{vibe_prompt}"
+
+Return exactly {count} real, verifiable books that perfectly capture this energy.
+Output ONLY a valid JSON array. No markdown fences. No text before or after.
+Schema:
+[
+  {{
+    "title": "Exact book title",
+    "author": "Author full name",
+    "reason": "One brilliant, atmospheric sentence explaining why it fits the requested vibe perfectly."
+  }}
+]
+"""
 
 class LLMService:
     """
@@ -779,6 +799,46 @@ def get_category_books(category: str, vibe_description: str, count: int = 5) ->
     logger.info("get_category_books: %d books returned for '%s'", len(valid_books), category)
     return valid_books
 
+def get_vibe_recommendations(vibe_prompt: str, count: int = 3) -> list:
+    """
+    Generate a list of real books based on a highly specific natural-language vibe.
+    Returns a list of dicts: [{"title": ..., "author": ..., "reason": ...}, ...]
+    """
+    if not llm_service.is_available():
+        logger.warning("get_vibe_recommendations: no LLM configured")
+        return []
+
+    prompt = PromptTemplates.get_vibe_check_prompt(vibe_prompt, count)
+    
+    # Utilizing the existing LLMService pipeline
+    raw = llm_service.generate_text(
+        prompt,
+        max_tokens=llm_service.config.get('category_books_max_tokens', 600),
+    )
+
+    if not raw:
+        logger.error("get_vibe_recommendations: LLM returned None")
+        return []
+
+    # Use the file's existing robust JSON extractor
+    parsed = _extract_json(raw)
+
+    if not isinstance(parsed, list):
+        logger.error("get_vibe_recommendations: expected JSON array")
+        return []
+
+    # Validate and clean the output
+    valid_books = []
+    for item in parsed:
+        if isinstance(item, dict) and "title" in item and "author" in item:
+            valid_books.append({
+                "title": item["title"],
+                "author": item["author"],
+                "reason": item.get("reason", "This perfectly matches your requested vibe."),
+            })
+
+    logger.info("get_vibe_recommendations: %d books returned for vibe '%s'", len(valid_books), vibe_prompt)
+    return valid_books
 
 @cache_mood_tags
 def get_book_mood_tags_safe(title: str, author: str = "") -> list:
@@ -902,3 +962,96 @@ def generate_chat_response(user_message: str, conversation_history: list = []) -
     ]
     import random
     return random.choice(variations)
+
+# =========================================================================
+# VIBE-CHECK ENDPOINT LOGIC
+# Implements strict validation and caching per PR #903 feedback.
+# =========================================================================
+
+import hashlib
+from cache_service import cache_service
+
+def generate_vibe_check(raw_vibe_prompt: str, raw_count: int = 5) -> Optional[dict]:
+    """
+    Generate an AI vibe check with strict server-side validation and caching.
+    
+    Args:
+        raw_vibe_prompt: The raw emotional/vibe query from the user.
+        raw_count: The requested number of book recommendations.
+        
+    Returns:
+        Dict containing the AI response and recommendations, or None on failure.
+    """
+    
+    # ---------------------------------------------------------
+    # 1. VALIDATION: Clamp the Count (Security & Cost Control)
+    # ---------------------------------------------------------
+    try:
+        count = int(raw_count)
+        if count < 1:
+            count = 1
+        elif count > 10:  # Hard limit to prevent API abuse
+            count = 10
+    except (ValueError, TypeError):
+        count = 5  # Safe default
+
+    # ---------------------------------------------------------
+    # 2. VALIDATION: Vibe Prompt Length (Prompt Injection Prevention)
+    # ---------------------------------------------------------
+    if not raw_vibe_prompt or not isinstance(raw_vibe_prompt, str):
+        logger.warning("generate_vibe_check: Empty or invalid prompt received.")
+        return None
+        
+    safe_vibe_prompt = raw_vibe_prompt.strip()
+    
+    # Restrict prompt to 400 characters to prevent huge payload bills
+    if len(safe_vibe_prompt) > 400:
+        logger.warning(f"generate_vibe_check: Prompt truncated from {len(safe_vibe_prompt)} chars.")
+        safe_vibe_prompt = safe_vibe_prompt[:400]
+
+    # ---------------------------------------------------------
+    # 3. CACHING: Prevent Expensive Redundant LLM Calls
+    # ---------------------------------------------------------
+    # Create a deterministic hash of the sanitized inputs
+    prompt_hash = hashlib.md5(f"{safe_vibe_prompt}_{count}".encode('utf-8')).hexdigest()
+    cache_key = f"vibe_check_{prompt_hash}"
+    
+    cached_result = cache_service.get(cache_key)
+    if cached_result:
+        logger.info(f"generate_vibe_check: Serving cached response for key {cache_key}")
+        return cached_result
+
+    # ---------------------------------------------------------
+    # 4. LLM GENERATION
+    # ---------------------------------------------------------
+    if not llm_service.is_available():
+        logger.error("generate_vibe_check: No LLM service available.")
+        return None
+
+    # Construct a specific prompt for the vibe check
+    system_prompt = (
+        f"You are a literary vibe-checker. The user has described their current mood: '{safe_vibe_prompt}'. "
+        f"Give them a short, poetic reading diagnosis (under 50 words) and recommend exactly {count} books "
+        "that perfectly match this specific emotional frequency. Format your response clearly."
+    )
+
+    try:
+        response_text = llm_service.generate_text(
+            prompt=system_prompt,
+            max_tokens=llm_service.config.get('recommendation_max_tokens', 250)
+        )
+        
+        if response_text:
+            result = {
+                "vibe_diagnosis": response_text,
+                "sanitized_query": safe_vibe_prompt,
+                "count_returned": count
+            }
+            
+            # Cache the successful result for 24 hours
+            cache_service.set(cache_key, result, timeout=86400)
+            return result
+            
+    except Exception as e:
+        logger.error(f"generate_vibe_check failed: {str(e)}", exc_info=True)
+        return None
@@ -42,7 +42,7 @@
 
 # Environment variables are now loaded centrally in backend/config.py
 from config import app_config, setup_logging, validate_required_env_vars
-from ai_service import generate_book_note, get_ai_recommendations, get_category_books, get_book_mood_tags_safe, generate_chat_response, llm_service
+from ai_service import generate_book_note, get_ai_recommendations, get_category_books, get_book_mood_tags_safe, generate_chat_response, llm_service, get_vibe_recommendations
 from models import db, User, Book, ShelfItem, BookNote, ReadingGoal, ReadingStats, Collection, CollectionItem, PriceHistory, PriceAlert, Review, register_user, login_user
 from price_tracker import get_price_tracker
 from cache_service import cache_service
@@ -912,6 +912,45 @@ def handle_mood_search():
         return internal_error(str(e))
 
 
+@app.route('/api/v1/vibe-check', methods=['POST'])
+@rate_limit('vibe_check')
+def handle_vibe_check():
+    """
+    Generate hyper-personalized book recommendations based on a specific vibe.
+    """
+    try:
+        data = request.get_json()
+
+        # Safely validate input
+        if not data or 'vibe_prompt' not in data:
+            return missing_fields_error(["vibe_prompt"])
+
+        vibe_prompt = data['vibe_prompt']
+        count = data.get('count', 3)
+
+        # Call the Gemini pipeline
+        books = get_vibe_recommendations(
+            vibe_prompt=vibe_prompt,
+            count=count
+        )
+
+        if not books:
+            return service_unavailable_error(
+                "Could not generate vibe recommendations right now. Please try again shortly."
+            )
+
+        return success_response(
+            data={
+                "vibe_prompt": vibe_prompt,
+                "books": books,
+            }
+        )
+
+    except Exception as e:
+        logger.error(f"Error in handle_vibe_check: {str(e)}", exc_info=True)
+        return internal_error(str(e))
+
+
 @app.route('/api/v1/category-books', methods=['POST'])
 @limiter.limit("10 per minute")
 def handle_category_books():