Update app.py

app.py

@@ -10,7 +10,7 @@ import praw
 import random
 import threading
 import nest_asyncio
-from datetime import datetime, date
+from datetime import datetime, timedelta
 from email.mime.text import MIMEText
 from email.mime.multipart import MIMEMultipart
 from openai import OpenAI
@@ -18,7 +18,6 @@ from playwright.async_api import async_playwright
 from apscheduler.schedulers.background import BackgroundScheduler
 from upstash_redis import Redis as UpstashRedis
 
-# إعدادات النظام
 os.environ["TZ"] = "UTC"
 nest_asyncio.apply()
 
@@ -26,356 +25,971 @@ nest_asyncio.apply()
 # 1. إعدادات البيئة والمفاتيح
 # ==========================================
 
-NVIDIA_API_KEY = os.getenv("NVIDIA_API_KEY")
-GITHUB_TOKEN = os.getenv("GITHUB_TOKEN")
-
-# Reddit Keys
-REDDIT_CLIENT_ID = os.getenv("REDDIT_CLIENT_ID")
+NVIDIA_API_KEY       = os.getenv("NVIDIA_API_KEY")
+GITHUB_TOKEN         = os.getenv("GITHUB_TOKEN")
+REDDIT_CLIENT_ID     = os.getenv("REDDIT_CLIENT_ID")
 REDDIT_CLIENT_SECRET = os.getenv("REDDIT_CLIENT_SECRET")
-REDDIT_USERNAME = os.getenv("REDDIT_USERNAME")
-REDDIT_PASSWORD = os.getenv("REDDIT_PASSWORD")
-
-# Email Keys
-SMTP_EMAIL = os.getenv("SMTP_EMAIL")
-SMTP_PASSWORD = os.getenv("SMTP_PASSWORD")
-SMTP_SERVER = "smtp.gmail.com"
-SMTP_PORT = 587
-
-# Upstash Redis Keys
-UPSTASH_REDIS_REST_URL = os.getenv("UPSTASH_REDIS_REST_URL")
+REDDIT_USERNAME      = os.getenv("REDDIT_USERNAME")
+REDDIT_PASSWORD      = os.getenv("REDDIT_PASSWORD")
+SMTP_EMAIL           = os.getenv("SMTP_EMAIL")
+SMTP_PASSWORD        = os.getenv("SMTP_PASSWORD")
+DEVTO_API_KEY        = os.getenv("DEVTO_API_KEY")           # اختياري
+DISCORD_BOT_TOKEN    = os.getenv("DISCORD_BOT_TOKEN")       # اختياري
+DISCORD_GUILD_IDS    = os.getenv("DISCORD_GUILD_IDS", "")   # معرفات السيرفرات (مفصولة بفاصلة)
+SMTP_SERVER  = "smtp.gmail.com"
+SMTP_PORT    = 587
+
+SITE_URL = "https://www.orgteh.com"
+
+UPSTASH_REDIS_REST_URL   = os.getenv("UPSTASH_REDIS_REST_URL")
 UPSTASH_REDIS_REST_TOKEN = os.getenv("UPSTASH_REDIS_REST_TOKEN")
 
-TELEGRAM_CONTACT = "@Nexusai97"
-
-# الحدود اليومية
+# حدود يومية للحماية من الحظر
 LIMITS = {
-    "reddit_daily_comments": 50,
-    "email_daily_send": 300,
-    "github_daily": 999999
+    "reddit_daily":  30,
+    "github_daily":  80,
+    "email_daily":  200,
+    "devto_daily":   25,
+    "hn_daily":      40,
+    "discord_daily": 20,
 }
 
 # ==========================================
-# 2. مدير التخزين والذاكرة (Iron Memory Manager)
+# 2. نظام الذاكرة الحديدية (Iron Memory)
 # ==========================================
 
-class LogManager:
+class IronMemory:
+    """
+    يحفظ كل URL / معرف مستخدم تم التعامل معه بشكل دائم.
+    يعمل على طبقتين: ذاكرة محلية (سريعة) + Upstash Redis (سحابية / دائمة).
+    لن نعود لأي شخص أو رابط تم حفظه أبداً حتى بعد إعادة التشغيل.
+    """
+    SET_KEY  = "orgteh:processed"
+    LOGS_KEY = "orgteh:logs"
+    USERS_KEY = "orgteh:users"        # مجموعة معرفات المستخدمين تحديداً
+
     def __init__(self):
+        self._local: set = set()
+        self._users: set = set()
+        self.redis = None
         self.enabled = False
         if UPSTASH_REDIS_REST_URL and UPSTASH_REDIS_REST_TOKEN:
             try:
                 self.redis = UpstashRedis(url=UPSTASH_REDIS_REST_URL, token=UPSTASH_REDIS_REST_TOKEN)
                 self.redis.ping()
                 self.enabled = True
-                print("✅ Connected to Upstash Redis (Iron Memory Active)")
+                print("✅ Iron Memory: Upstash Redis متصل — الذاكرة الدائمة نشطة")
             except Exception as e:
-                print(f"❌ Upstash Connection Error: {e}")
+                print(f"❌ Upstash خطأ: {e} — سيعمل النظام بذاكرة مؤقتة فقط")
         else:
-            print("⚠️ Upstash Credentials missing. Memory will be temporary only!")
+            print("⚠️ بيانات Upstash غير موجودة — الذاكرة مؤقتة فقط")
 
-    def is_processed(self, url):
-        """فحص هل تم التعامل مع الرابط سابقاً (سحابياً)"""
-        # أولاً نفحص الذاكرة المحلية للسرعة
-        if url in LOCAL_STATS['processed_urls']:
+    # --- فحص ---
+    def seen(self, key: str) -> bool:
+        if key in self._local:
             return True
-        
-        # ثانياً نفحص السحابة (للأمان بعد إعادة التشغيل)
         if self.enabled:
             try:
-                # نستخدم أمر sismember للتحقق من وجود الرابط في المجموعة
-                return self.redis.sismember("nexus_processed_set", url) == 1
+                return bool(self.redis.sismember(self.SET_KEY, key))
             except:
-                return False
+                pass
         return False
 
-    def mark_processed(self, url):
-        """تسجيل الرابط بأنه تم التعامل معه"""
-        LOCAL_STATS['processed_urls'].append(url)
+    def user_seen(self, user_id: str) -> bool:
+        if user_id in self._users:
+            return True
         if self.enabled:
             try:
-                # إضافة الرابط لمجموعة الروابط المعالجة
-                self.redis.sadd("nexus_processed_set", url)
-            except: pass
+                return bool(self.redis.sismember(self.USERS_KEY, user_id))
+            except:
+                pass
+        return False
 
-    def load_logs(self):
-        if not self.enabled: return []
-        try:
-            raw_logs = self.redis.lrange("nexus_logs", 0, 99)
-            parsed_logs = []
-            for log in raw_logs:
-                if isinstance(log, str):
-                    try: parsed_logs.append(json.loads(log))
-                    except: pass
-                else: parsed_logs.append(log)
-            return parsed_logs
-        except: return []
-
-    def save_log_entry(self, platform, title, url, original_text, our_reply):
+    # --- تسجيل ---
+    def mark(self, key: str):
+        self._local.add(key)
+        if self.enabled:
+            try:
+                self.redis.sadd(self.SET_KEY, key)
+            except:
+                pass
+
+    def mark_user(self, user_id: str):
+        self._users.add(user_id)
+        if self.enabled:
+            try:
+                self.redis.sadd(self.USERS_KEY, user_id)
+            except:
+                pass
+
+    # --- السجلات ---
+    def log(self, platform, title, url, snippet, reply):
         entry = {
-            "timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
+            "ts":       datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
             "platform": platform,
-            "title": title,
-            "url": url,
-            "original_text": original_text[:500] + "..." if original_text else "No content",
-            "our_reply": our_reply
+            "title":    title,
+            "url":      url,
+            "snippet":  (snippet or "")[:400],
+            "reply":    reply,
         }
-        
         LOCAL_LOGS.insert(0, entry)
-        if len(LOCAL_LOGS) > 100: LOCAL_LOGS.pop()
-        
+        if len(LOCAL_LOGS) > 200:
+            LOCAL_LOGS.pop()
         if self.enabled:
             try:
-                self.redis.lpush("nexus_logs", json.dumps(entry))
-                self.redis.ltrim("nexus_logs", 0, 499)
-            except: pass
+                self.redis.lpush(self.LOGS_KEY, json.dumps(entry))
+                self.redis.ltrim(self.LOGS_KEY, 0, 999)
+            except:
+                pass
         return entry
 
-log_manager = LogManager()
+    def load_logs(self):
+        if self.enabled:
+            try:
+                raw = self.redis.lrange(self.LOGS_KEY, 0, 199)
+                result = []
+                for item in raw:
+                    try:
+                        result.append(json.loads(item) if isinstance(item, str) else item)
+                    except:
+                        pass
+                return result
+            except:
+                pass
+        return LOCAL_LOGS[:]
+
 
-LOCAL_STATS = {"reddit": 0, "github": 0, "email": 0, "processed_urls": []}
+memory = IronMemory()
 LOCAL_LOGS = []
+LOCAL_STATS = {"reddit": 0, "github": 0, "email": 0, "devto": 0, "hn": 0, "discord": 0}
 
 # ==========================================
-# 3. الذكاء الاصطناعي (AI Core)
+# 3. محرك الذكاء الاصطناعي
 # ==========================================
 
 client = OpenAI(base_url="https://integrate.api.nvidia.com/v1", api_key=NVIDIA_API_KEY)
 
-def ask_ai(prompt, system_role="You are a helpful assistant."):
+def ask_ai(prompt: str, system: str = "You are a helpful assistant.", temp=0.5) -> str | None:
     try:
-        completion = client.chat.completions.create(
+        resp = client.chat.completions.create(
             model="meta/llama-3.3-70b-instruct",
-            messages=[{"role": "system", "content": system_role}, {"role": "user", "content": prompt}],
-            temperature=0.3, max_tokens=800
+            messages=[{"role": "system", "content": system}, {"role": "user", "content": prompt}],
+            temperature=temp, max_tokens=900
         )
-        return completion.choices[0].message.content
+        return resp.choices[0].message.content.strip()
     except Exception as e:
         print(f"AI Error: {e}")
         return None
 
-def generate_polyglot_keywords():
-    resp = ask_ai("Generate 6 search keywords for finding people needing API, Chatbots, or AI services. Mix English, Arabic, Turkish, Chinese. Output ONLY keywords separated by commas.")
-    if resp: return [k.strip() for k in resp.split(",") if len(k) > 3]
-    return ["cheap api", "برمجة شات بوت", "openai alternative", "ai integration"]
-
-def generate_localized_reply(content, platform):
-    prompt = (
-        f"Analyze content from {platform}:\n'''{content[:800]}'''\n"
-        "1. DETECT language.\n"
-        "2. WRITE a short, direct reply in the SAME LANGUAGE.\n"
-        "3. PERSONA: You are the OWNER (Individual Developer). Use 'I' (not 'We').\n"
-        "4. TONE: Direct, Confident, Humble. (No corporate jargon).\n"
-        f"5. TEMPLATE (Adapt to language): 'Okay, I own an API [or Chatbot service] with multiple models and high performance at a cheap price. Message me on Telegram {TELEGRAM_CONTACT}.'\n"
-        "6. If the user asks for Chatbot, say 'I own a chatbot service'. If API, say 'I own an API'.\n"
-        "Output ONLY the reply text."
-    )
-    return ask_ai(prompt)
-
-def generate_cold_email_content(website_content, company_name):
-    prompt = (
-        f"Company: {company_name}\nSnippet: {website_content[:500]}\n"
-        "DETECT language. Write JSON Cold Email: { 'subject': '...', 'body': '...' }. "
-        "Tone: Personal owner offering help. Use 'I'. "
-        "Offer Nexus AI services (API & Chatbots). Output strict JSON."
-    )
-    resp = ask_ai(prompt)
-    try:
-        json_match = re.search(r'\{.*\}', resp, re.DOTALL)
-        if json_match: return json.loads(json_match.group(0))
-    except: pass
+# ==========================================
+# 4. نظام كلمات البحث الذكي — تلقائي 100%
+# ==========================================
+# طبقتان:
+#   1. FALLBACK_KEYWORDS  — كلمات ثابتة لضمان العمل دائماً
+#   2. كل KEYWORD_REFRESH_MINUTES يطلب النموذج توليد كلمات جديدة بنفسه
+#      بكل اللغات الممكنة بناءً على ما يعرفه عن Orgteh وسوق AI API
+
+# ── كلمات ثابتة (Fallback) — تُستخدم عند بدء التشغيل أو فشل التوليد ──
+FALLBACK_KEYWORDS: list[str] = [
+    # EN
+    "llm api", "ai api", "chatbot api", "language model api",
+    "integrate llm", "openai api", "deepseek api", "mistral api",
+    "llama api", "ai api developer", "chat completions api",
+    # AR
+    "api ذكاء اصطناعي", "دمج نموذج لغوي", "api شات بوت",
+    # TR
+    "yapay zeka api", "llm api entegrasyonu",
+    # FR
+    "api intelligence artificielle", "intégrer llm api",
+    # ES
+    "api inteligencia artificial", "integrar llm api",
+    # DE
+    "ki api", "sprachmodell api",
+    # PT
+    "api inteligência artificial", "chatbot api integração",
+    # RU
+    "языковая модель api", "llm api интеграция",
+    # ZH
+    "大语言模型 api", "ai api 接入",
+    # JA
+    "llm api 統合", "言語モデル api",
+    # KO
+    "llm api 통합", "챗봇 api 연동",
+    # HI
+    "ai api इंटीग्रेशन", "chatbot api बनाना",
+    # ID
+    "api kecerdasan buatan", "chatbot api developer",
+]
+
+KEYWORD_REFRESH_MINUTES = 90   # كل 90 دقيقة يطلب كلمات جديدة
+_keyword_cache: list[str] = list(FALLBACK_KEYWORDS)
+_last_keyword_refresh: float = 0.0
+_keyword_lock = threading.Lock()
+
+
+def _ai_generate_keywords() -> list[str]:
+    """
+    يطلب من النموذج توليد كلمات بحث جديدة بنفسه —
+    بكل اللغات والزوايا الممكنة لإيجاد المطورين المحتاجين لـ AI API.
+    """
+    prompt = """You are helping market Orgteh (orgteh.com) — an AI API service for developers.
+Orgteh provides: OpenAI-compatible LLM API, multiple models (DeepSeek, Mistral, Llama, Kimi, Gemma), cheap pricing.
+Target: developers and startups who want to USE or ACCESS an AI language model API for their projects.
+
+Generate 40 short search keywords (2-5 words each) to find these people on GitHub, Reddit, forums, and the web.
+- Cover MANY languages: English, Arabic, Turkish, French, Spanish, German, Portuguese, Russian, Chinese, Japanese, Korean, Hindi, Indonesian, Italian, Polish, Dutch, Vietnamese, Thai, and others you know.
+- Include different angles: building chatbots, integrating LLMs, looking for API access, seeking cheaper alternatives, starting AI projects.
+- Keep each keyword SHORT and natural (as someone would actually search or post).
+- Output ONLY a JSON array of strings, no explanation:
+["keyword1", "keyword2", ...]"""
+
+    resp = ask_ai(prompt, temp=0.8)
+    if resp:
+        try:
+            m = re.search(r'\[.*\]', resp, re.DOTALL)
+            if m:
+                keywords = json.loads(m.group(0))
+                valid = [k.strip() for k in keywords if isinstance(k, str) and 2 < len(k.strip()) < 60]
+                if len(valid) >= 10:
+                    print(f"🔄 Keywords refreshed: {len(valid)} new keywords generated by AI")
+                    return valid
+        except Exception as e:
+            print(f"Keyword generation parse error: {e}")
+    return []
+
+
+def refresh_keywords_if_needed():
+    """يجدد كلمات البحث تلقائياً كل KEYWORD_REFRESH_MINUTES."""
+    global _keyword_cache, _last_keyword_refresh
+    now = time.time()
+    with _keyword_lock:
+        if now - _last_keyword_refresh > KEYWORD_REFRESH_MINUTES * 60:
+            new_kws = _ai_generate_keywords()
+            if new_kws:
+                # نمزج الكلمات الجديدة مع الثابتة لضمان التغطية
+                _keyword_cache = list(set(FALLBACK_KEYWORDS + new_kws))
+                _last_keyword_refresh = now
+
+
+def _keyword_refresh_loop():
+    """خيط مستقل يجدد الكلمات بشكل دوري."""
+    # توليد فوري عند البدء
+    refresh_keywords_if_needed()
+    while True:
+        time.sleep(KEYWORD_REFRESH_MINUTES * 60)
+        refresh_keywords_if_needed()
+
+
+def get_random_keywords(n: int = 3) -> list[str]:
+    """يختار n كلمات عشوائية من الكاش الحالي (ثابتة + مولّدة)."""
+    with _keyword_lock:
+        pool = _keyword_cache if _keyword_cache else FALLBACK_KEYWORDS
+    return random.sample(pool, min(n, len(pool)))
+
+# ==========================================
+# 5. مولّد الردود الذكي
+# ==========================================
+
+SITE_CONTEXT = """
+Orgteh (orgteh.com) is an AI API service offering:
+- OpenAI-compatible endpoint: https://orgteh.com/v1/chat/completions
+- Multiple models: DeepSeek-R1, Kimi (256K context), Mistral, Llama-3B, Gemma
+- API keys start with "Orgteh_"
+- Features: GPU-accelerated, sub-second response, enterprise security, streaming support
+- Extra tools: web scraper, OCR, embeddings, finance/news aggregator, AI Code Builder
+- Cheaper than OpenAI with the same integration format (drop-in replacement, zero code changes)
+- Registration: orgteh.com/register
+"""
+
+REPLY_STYLES = [
+    "casual_developer",   # مطور يشارك تجربته
+    "helpful_peer",       # زميل يساعد
+    "been_there",         # شخص مر بنفس المشكلة
+]
+
+def ai_judge(text: str) -> bool:
+    """
+    النموذج نفسه يحكم: هل هذا الشخص يحتاج فعلاً لـ API نماذج ذكاء اصطناعي؟
+    لا كلمات مفتاحية — فهم النية الحقيقية فقط.
+    """
+    prompt = f"""Read this post/issue carefully:
+\"\"\"
+{text[:600]}
+\"\"\"
+
+Answer with ONE word only — YES or NO:
+Is this person actively trying to USE or ACCESS an AI language model API
+(to build something, integrate it, or get API access for a real project)?
+
+Rules:
+- YES: they are building/integrating/seeking API access to an AI model for a concrete use case.
+- NO: discussing AI concepts/news/opinions, reporting unrelated bugs, academic questions, or anything not about actually using a model API.
+- If unsure → NO.
+
+Answer:"""
+
+    resp = ask_ai(prompt, temp=0.0)
+    if resp:
+        return resp.strip().upper().startswith("YES")
+    return False
+
+def generate_reply(content: str, platform: str) -> str | None:
+    """يولد رد يبدو كمستخدم طبيعي اكتشف orgteh وأراد مشاركته."""
+    style = random.choice(REPLY_STYLES)
+    prompt = f"""
+You are a real developer who uses Orgteh (orgteh.com) for AI API access.
+You are browsing {platform} and found this post/issue that matches your past experience.
+You want to genuinely help by sharing what worked for you.
+
+CONTEXT ABOUT ORGTEH:
+{SITE_CONTEXT}
+
+POST/ISSUE CONTENT:
+\"\"\"
+{content[:700]}
+\"\"\"
+
+YOUR PERSONA STYLE TODAY: {style}
+
+STRICT RULES:
+1. Detect the language of the post and reply in THE SAME LANGUAGE.
+2. Write as a REAL PERSON, NOT a company. Use "I" not "we".
+3. Sound natural and conversational. NOT salesy or promotional.
+4. Keep it SHORT (3-6 sentences max).
+5. Mention orgteh.com naturally as "I've been using orgteh.com" or "stumbled on orgteh.com" etc.
+6. Only reference features that are actually relevant to their specific problem.
+7. DO NOT start with "Hi" or "Hello" or generic openers.
+8. DO NOT use bullet points or formatting.
+9. Output ONLY the reply text, nothing else.
+
+Examples of good natural openers (adapt to language/situation):
+- "Been using orgteh.com's API for this exact use case, works well with Python..."
+- "I built something similar, ended up using orgteh.com — they have an OpenAI-compatible endpoint so..."
+- "For LLM API access orgteh.com has been solid for me, multiple models under one endpoint..."
+- "جربت orgteh.com لنفس الغرض، API متوافق مع OpenAI ويشتغل مباشرة..."
+- "عندي تجربة مع هذا، استخدمت orgteh.com وفيه عدة نماذج بـ endpoint موحد..."
+"""
+    return ask_ai(prompt, temp=0.7)
+
+def generate_email(company: str, snippet: str) -> dict | None:
+    """يولد إيميل بارد احترافي وشخصي لاستهداف الشركات والمطورين."""
+    prompt = f"""
+You are a developer who uses Orgteh API and wants to recommend it to a company/developer who seems to need AI API services.
+
+COMPANY/DEV NAME: {company}
+WHAT THEY DO (from their website): {snippet[:500]}
+
+CONTEXT ABOUT ORGTEH:
+{SITE_CONTEXT}
+
+Write a cold email as a fellow developer (not a company rep).
+- Detect language from the snippet and write in the SAME LANGUAGE.
+- Subject: short, specific to their use case.
+- Body: personal, 4-6 sentences. Reference what they do specifically.
+- Sound like you're sharing a tool that helped you, not selling.
+- Include orgteh.com naturally.
+- Use "I" not "we".
+
+Output STRICT JSON only:
+{{"subject": "...", "body": "..."}}
+"""
+    resp = ask_ai(prompt, temp=0.6)
+    if resp:
+        try:
+            m = re.search(r'\{.*\}', resp, re.DOTALL)
+            if m:
+                return json.loads(m.group(0))
+        except:
+            pass
     return None
 
 # ==========================================
-# 4. دوال التنفيذ (Executors)
+# 6. نظام مانع الحظر (Anti-Ban Engine)
 # ==========================================
 
-def send_smtp_email(to_email, subject, body):
-    if not SMTP_EMAIL or not SMTP_PASSWORD: return False, "No Config"
-    try:
-        msg = MIMEMultipart()
-        msg['From'] = SMTP_EMAIL
-        msg['To'] = to_email
-        msg['Subject'] = subject
-        msg.attach(MIMEText(body, 'plain'))
-        server = smtplib.SMTP(SMTP_SERVER, SMTP_PORT)
-        server.starttls()
-        server.login(SMTP_EMAIL, SMTP_PASSWORD)
-        server.send_message(msg)
-        server.quit()
-        return True, "Sent"
-    except Exception as e: return False, str(e)
+class AntiBan:
+    """
+    يتحكم في توقيت الإرسال لكل منصة لتجنب الحظر.
+    يستخدم تأخيرات عشوائية وحدوداً يومية صارمة.
+    """
+    def __init__(self):
+        self._day      = datetime.utcnow().date()
+        self._counts   = {k: 0 for k in LIMITS}
+        self._lock     = threading.Lock()
+
+    def _reset_if_new_day(self):
+        today = datetime.utcnow().date()
+        if today != self._day:
+            self._day    = today
+            self._counts = {k: 0 for k in LIMITS}
+
+    def can_act(self, platform: str) -> bool:
+        key = f"{platform}_daily"
+        with self._lock:
+            self._reset_if_new_day()
+            return self._counts.get(key, 0) < LIMITS.get(key, 999)
+
+    def record(self, platform: str):
+        key = f"{platform}_daily"
+        with self._lock:
+            self._counts[key] = self._counts.get(key, 0) + 1
+            LOCAL_STATS[platform] = LOCAL_STATS.get(platform, 0) + 1
+
+    @staticmethod
+    def human_delay(base=15, jitter=20):
+        """تأخير عشوائي يشبه السلوك البشري."""
+        delay = base + random.uniform(0, jitter)
+        time.sleep(delay)
+
+    @staticmethod
+    def micro_delay():
+        time.sleep(random.uniform(2, 6))
+
+
+antiban = AntiBan()
 
 # ==========================================
-# 5. الحلقات (The Loops) - Protected
+# 7. منصات البث - GitHub
 # ==========================================
 
-# --- GitHub Loop ---
 def github_loop():
+    headers = {"Authorization": f"token {GITHUB_TOKEN}", "Accept": "application/vnd.github.v3+json"}
     while True:
         try:
-            keywords = generate_polyglot_keywords()
-            headers = {"Authorization": f"token {GITHUB_TOKEN}", "Accept": "application/vnd.github.v3+json"}
-            kw = random.choice(keywords)
-            
-            resp = requests.get(f"https://api.github.com/search/issues?q={kw}+state:open&sort=updated", headers=headers)
-            if resp.status_code == 200:
-                for item in resp.json().get("items", [])[:3]:
-                    # 🛡️ الحماية الحديدية: فحص السحابة قبل العمل
-                    if log_manager.is_processed(item['html_url']): continue
-                    
-                    reply = generate_localized_reply(item['title'] + "\n" + (item['body'] or ""), "GitHub")
-                    if reply:
-                        post_resp = requests.post(item['url']+"/comments", headers=headers, json={"body": reply})
-                        if post_resp.status_code == 201:
-                            LOCAL_STATS["github"] += 1
-                            # 🛡️ تسجيل الرابط في السحابة
-                            log_manager.mark_processed(item['html_url'])
-                            log_manager.save_log_entry("GitHub", item['title'], item['html_url'], item['body'], reply)
-                            print(f"✅ GitHub Replied: {item['title'][:20]}")
-                            time.sleep(60)
-            time.sleep(120)
+            if not GITHUB_TOKEN:
+                time.sleep(600)
+                continue
+
+            keywords = get_random_keywords(2)
+            for kw in keywords:
+                if not antiban.can_act("github"):
+                    break
+
+                # البحث في Issues
+                resp = requests.get(
+                    f"https://api.github.com/search/issues"
+                    f"?q={requests.utils.quote(kw)}+state:open+type:issue"
+                    f"&sort=updated&per_page=5",
+                    headers=headers, timeout=15
+                )
+                if resp.status_code != 200:
+                    if resp.status_code == 403:
+                        time.sleep(300)
+                    break
+
+                for item in resp.json().get("items", []):
+                    url       = item["html_url"]
+                    body_text = (item.get("body") or "")
+                    full_text = item["title"] + " " + body_text
+
+                    if memory.seen(url):
+                        continue
+
+                    user_id = f"gh:{item.get('user', {}).get('login', '')}"
+                    if memory.user_seen(user_id):
+                        continue
+
+                    # النموذج يحكم على النية — لا شروط كلمية
+                    if not ai_judge(full_text):
+                        print(f"  ⏭ GitHub skip (not relevant): {item['title'][:40]}")
+                        continue
+
+                    reply = generate_reply(full_text, "GitHub")
+                    if not reply:
+                        continue
+
+                    post = requests.post(
+                        item["url"] + "/comments",
+                        headers=headers,
+                        json={"body": reply},
+                        timeout=10
+                    )
+                    if post.status_code == 201:
+                        memory.mark(url)
+                        memory.mark_user(user_id)
+                        memory.log("GitHub", item["title"], url, body_text, reply)
+                        antiban.record("github")
+                        print(f"✅ GitHub: {item['title'][:50]}")
+                        antiban.human_delay(60, 60)
+
+            time.sleep(random.randint(90, 180))
+
         except Exception as e:
+            print(f"GitHub Error: {e}")
             time.sleep(300)
 
-# --- Email Loop ---
-async def email_hunter_process():
+# ==========================================
+# 8. منصات البث - Reddit
+# ==========================================
+
+# سابريدتات مناسبة فقط
+TARGET_SUBREDDITS = [
+    "MachineLearning", "LocalLLaMA", "learnmachinelearning",
+    "artificial", "ChatGPT", "OpenAI", "SideProject",
+    "startups", "webdev", "learnprogramming", "Python",
+    "programming", "ArtificialIntelligence",
+]
+
+def reddit_loop():
+    while True:
+        try:
+            if not REDDIT_CLIENT_ID or not antiban.can_act("reddit"):
+                time.sleep(600)
+                continue
+
+            reddit = praw.Reddit(
+                client_id=REDDIT_CLIENT_ID, client_secret=REDDIT_CLIENT_SECRET,
+                username=REDDIT_USERNAME, password=REDDIT_PASSWORD,
+                user_agent="orgteh-community-bot/1.0"
+            )
+
+            kw = random.choice(get_random_keywords(1))
+            sub = random.choice(TARGET_SUBREDDITS)
+
+            for post in reddit.subreddit(sub).search(kw, limit=5, sort="new"):
+                if not antiban.can_act("reddit"):
+                    break
+
+                full_text = post.title + " " + (post.selftext or "")
+                if not ai_judge(full_text):
+                    continue
+
+                url     = post.url
+                user_id = f"rd:{post.author}"
+                if memory.seen(url) or memory.user_seen(user_id):
+                    continue
+
+                reply = generate_reply(full_text, "Reddit")
+                if not reply:
+                    continue
+
+                post.reply(reply)
+                memory.mark(url)
+                memory.mark_user(user_id)
+                memory.log("Reddit", post.title, url, post.selftext or "", reply)
+                antiban.record("reddit")
+                print(f"✅ Reddit: {post.title[:50]}")
+                antiban.human_delay(120, 180)
+                break  # رد واحد لكل دورة
+
+        except Exception as e:
+            print(f"Reddit Error: {e}")
+        time.sleep(random.randint(900, 1500))
+
+# ==========================================
+# 9. منصات البث - Hacker News
+# ==========================================
+
+def hn_loop():
+    """Hacker News Ask HN / Show HN عبر Algolia API المجاني."""
+    base = "https://hn.algolia.com/api/v1/search"
+    while True:
+        try:
+            if not antiban.can_act("hn"):
+                time.sleep(600)
+                continue
+
+            kw = random.choice(SEED_KEYWORDS_EN[:8])
+            resp = requests.get(base, params={
+                "query": kw, "tags": "ask_hn,story",
+                "hitsPerPage": 5, "numericFilters": "points>5"
+            }, timeout=15)
+
+            if resp.status_code != 200:
+                time.sleep(300)
+                continue
+
+            for hit in resp.json().get("hits", []):
+                story_id  = str(hit.get("objectID"))
+                full_text = (hit.get("title","") + " " + (hit.get("story_text") or ""))
+
+                if not ai_judge(full_text):
+                    continue
+
+                url     = f"https://news.ycombinator.com/item?id={story_id}"
+                user_id = f"hn:{hit.get('author','')}"
+                if memory.seen(url) or memory.user_seen(user_id):
+                    continue
+
+                # HN لا يسمح بالتعليق عبر API — نسجل فقط للمراجعة اليدوية
+                # لكن نولد الرد للاطلاع
+                reply = generate_reply(full_text, "HackerNews")
+                if reply:
+                    memory.mark(url)
+                    memory.mark_user(user_id)
+                    memory.log("HackerNews (Manual)", hit.get("title",""), url, full_text[:300], reply)
+                    antiban.record("hn")
+                    print(f"📝 HN (للمراجعة): {hit.get('title','')[:50]}")
+                    antiban.micro_delay()
+
+        except Exception as e:
+            print(f"HN Error: {e}")
+        time.sleep(random.randint(600, 900))
+
+# ==========================================
+# 10. منصات البث - Dev.to
+# ==========================================
+
+def devto_loop():
+    """Dev.to API - التعليق على مقالات المطورين."""
+    headers = {"api-key": DEVTO_API_KEY or "", "Content-Type": "application/json"}
+    while True:
+        try:
+            if not DEVTO_API_KEY or not antiban.can_act("devto"):
+                time.sleep(1800)
+                continue
+
+            for tag in ["api", "llm", "openai", "chatbot", "artificialintelligence"]:
+                resp = requests.get(
+                    f"https://dev.to/api/articles?tag={tag}&top=1&per_page=5",
+                    timeout=10
+                )
+                if resp.status_code != 200:
+                    continue
+
+                for article in resp.json():
+                    full_text = article.get("title","") + " " + (article.get("description") or "")
+                    if not ai_judge(full_text):
+                        continue
+
+                    url     = article.get("url","")
+                    user_id = f"devto:{article.get('user',{}).get('username','')}"
+                    if not url or memory.seen(url) or memory.user_seen(user_id):
+                        continue
+
+                    reply = generate_reply(full_text, "Dev.to")
+                    if not reply:
+                        continue
+
+                    post = requests.post(
+                        f"https://dev.to/api/comments",
+                        headers=headers,
+                        json={"body_markdown": reply, "article_id": article["id"]},
+                        timeout=10
+                    )
+                    if post.status_code in (200, 201):
+                        memory.mark(url)
+                        memory.mark_user(user_id)
+                        memory.log("Dev.to", article["title"], url, full_text, reply)
+                        antiban.record("devto")
+                        print(f"✅ Dev.to: {article['title'][:50]}")
+                        antiban.human_delay(90, 90)
+                        break
+
+        except Exception as e:
+            print(f"Dev.to Error: {e}")
+        time.sleep(random.randint(1800, 2700))
+
+# ==========================================
+# 11. منصات البث - Discord (قراءة فقط → رد عبر DM)
+# ==========================================
+
+def discord_monitor_loop():
+    """
+    يراقب سيرفرات Discord المحددة ويبحث عن رسائل عن API.
+    يرسل DM للأشخاص المحتاجين (ليس رد عام للسيرفر).
+    يتطلب: DISCORD_BOT_TOKEN + DISCORD_GUILD_IDS
+    """
+    if not DISCORD_BOT_TOKEN:
+        return
+
+    headers = {
+        "Authorization": f"Bot {DISCORD_BOT_TOKEN}",
+        "Content-Type": "application/json"
+    }
+    guild_ids = [g.strip() for g in DISCORD_GUILD_IDS.split(",") if g.strip()]
+
+    while True:
+        try:
+            if not antiban.can_act("discord"):
+                time.sleep(3600)
+                continue
+
+            for guild_id in guild_ids:
+                # جلب قنوات السيرفر
+                ch_resp = requests.get(
+                    f"https://discord.com/api/v10/guilds/{guild_id}/channels",
+                    headers=headers, timeout=10
+                )
+                if ch_resp.status_code != 200:
+                    continue
+
+                text_channels = [c for c in ch_resp.json() if c.get("type") == 0]
+                for channel in text_channels[:5]:
+                    # جلب آخر الرسائل
+                    msg_resp = requests.get(
+                        f"https://discord.com/api/v10/channels/{channel['id']}/messages?limit=20",
+                        headers=headers, timeout=10
+                    )
+                    if msg_resp.status_code != 200:
+                        continue
+
+                    for msg in msg_resp.json():
+                        content = msg.get("content", "")
+                        msg_id  = msg.get("id", "")
+                        user    = msg.get("author", {})
+                        user_id = f"dc:{user.get('id','')}"
+
+                        if user.get("bot"):
+                            continue
+                        if not ai_judge(content):
+                            continue
+                        if memory.seen(msg_id) or memory.user_seen(user_id):
+                            continue
+
+                        # إرسال DM
+                        dm_resp = requests.post(
+                            "https://discord.com/api/v10/users/@me/channels",
+                            headers=headers,
+                            json={"recipient_id": user["id"]},
+                            timeout=10
+                        )
+                        if dm_resp.status_code != 200:
+                            continue
+
+                        dm_channel = dm_resp.json().get("id")
+                        reply = generate_reply(content, "Discord")
+                        if not reply:
+                            continue
+
+                        send = requests.post(
+                            f"https://discord.com/api/v10/channels/{dm_channel}/messages",
+                            headers=headers,
+                            json={"content": reply},
+                            timeout=10
+                        )
+                        if send.status_code in (200, 201):
+                            memory.mark(msg_id)
+                            memory.mark_user(user_id)
+                            memory.log("Discord DM", channel.get("name",""), msg_id, content, reply)
+                            antiban.record("discord")
+                            print(f"✅ Discord DM → {user.get('username','')}")
+                            antiban.human_delay(120, 120)
+                            break
+
+        except Exception as e:
+            print(f"Discord Error: {e}")
+        time.sleep(random.randint(3600, 5400))
+
+# ==========================================
+# 12. نظام الإيميل المتقدم
+# ==========================================
+
+def send_email(to: str, subject: str, body: str) -> tuple[bool, str]:
+    if not SMTP_EMAIL or not SMTP_PASSWORD:
+        return False, "SMTP غير مهيأ"
     try:
-        if LOCAL_STATS["email"] >= LIMITS["email_daily_send"]: return
-        kw = random.choice(generate_polyglot_keywords())
-        
-        async with async_playwright() as p:
-            browser = await p.chromium.launch(
-                headless=True,
-                args=['--no-sandbox', '--disable-setuid-sandbox', '--disable-dev-shm-usage', '--disable-gpu']
+        msg = MIMEMultipart()
+        msg["From"]    = SMTP_EMAIL
+        msg["To"]      = to
+        msg["Subject"] = subject
+        msg.attach(MIMEText(body, "plain", "utf-8"))
+        srv = smtplib.SMTP(SMTP_SERVER, SMTP_PORT)
+        srv.starttls()
+        srv.login(SMTP_EMAIL, SMTP_PASSWORD)
+        srv.send_message(msg)
+        srv.quit()
+        return True, "تم الإرسال"
+    except Exception as e:
+        return False, str(e)
+
+# استراتيجيات البحث عن الإيميلات والشركات الناشئة
+EMAIL_SEARCH_QUERIES = [
+    # يبني تطبيق على AI API
+    "site:github.io ai api project contact email",
+    "site:github.com llm api integration project contact",
+    '"llm api" developer project contact email',
+    '"ai api" developer "contact" OR "email" site:github.com',
+    # يبحث عن كيفية الاستخدام
+    '"how to use llm api" developer email',
+    '"integrate ai model" project developer contact',
+    # شركات ناشئة تبني على AI
+    "site:producthunt.com ai chatbot api developer 2024",
+    "site:producthunt.com llm powered app developer",
+    "startup ai api integration developer email contact",
+    "site:crunchbase.com ai startup llm api developer",
+    # مطورون مستقلون
+    "site:indie.hackers.com ai api llm project",
+    '"building with llm" developer email contact',
+    '"ai powered" app developer "reach me" OR "contact"',
+    # بحث عام
+    '"llm api" OR "ai api" developer startup email contact 2024',
+    '"chat completions" developer project contact email',
+]
+
+async def email_hunter():
+    """يبحث عن مطورين وشركات ناشئة تحتاج فعلاً لـ API ذكاء اصطناعي."""
+    if not antiban.can_act("email"):
+        return
+
+    query = random.choice(EMAIL_SEARCH_QUERIES)
+
+    async with async_playwright() as p:
+        browser = await p.chromium.launch(headless=True, args=[
+            "--no-sandbox", "--disable-setuid-sandbox",
+            "--disable-dev-shm-usage", "--disable-gpu"
+        ])
+        page = await browser.new_page()
+        await page.set_extra_http_headers({"Accept-Language": "en-US,en;q=0.9"})
+
+        try:
+            # البحث في DuckDuckGo
+            await page.goto(
+                f"https://duckduckgo.com/?q={requests.utils.quote(query)}&t=h_&ia=web",
+                timeout=25000
             )
-            page = await browser.new_page()
-            
-            try:
-                await page.goto(f"https://duckduckgo.com/?q={kw}+email+contact&t=h_&ia=web", timeout=20000)
-                links = await page.query_selector_all('a.result__a')
-                
-                extracted = []
-                for l in links[:5]:
-                    u = await l.get_attribute('href')
-                    t = await l.inner_text()
-                    if u and "facebook" not in u and "linkedin" not in u:
-                        extracted.append((u, t))
-                
-                for url, title in extracted:
-                    if LOCAL_STATS["email"] >= LIMITS["email_daily_send"]: break
-                    # 🛡️ الحماية الحديدية
-                    if log_manager.is_processed(url): continue
+            await asyncio.sleep(3)
+
+            links = await page.query_selector_all("a.result__a")
+            targets = []
+            for l in links[:8]:
+                href = await l.get_attribute("href")
+                text = await l.inner_text()
+                if href and all(x not in href for x in ["facebook", "linkedin", "twitter", "youtube"]):
+                    targets.append((href, text.strip()))
+
+            for url, title in targets:
+                if not antiban.can_act("email"):
+                    break
+                if memory.seen(url):
+                    continue
+
+                try:
+                    await page.goto(url, timeout=20000)
+                    await asyncio.sleep(2)
+                    content = await page.content()
+
+                    # استخراج الإيميلات
+                    emails = list(set(re.findall(
+                        r"[a-zA-Z0-9._%+\-]+@[a-zA-Z0-9.\-]+\.[a-zA-Z]{2,}", content
+                    )))
+                    emails = [e for e in emails if not any(
+                        e.endswith(x) for x in [".png", ".jpg", ".js", ".css", ".svg", ".gif"]
+                    ) and "example" not in e and "noreply" not in e]
+
+                    if not emails:
+                        continue
+
+                    # التحقق من الصلة بالموضوع
+                    page_text = re.sub(r"<[^>]+>", " ", content)
+                    if not ai_judge(page_text[:2000]):
+                        continue
+
+                    target_email = emails[0]
+                    email_id     = f"email:{target_email}"
+                    if memory.user_seen(email_id):
+                        continue
+
+                    # توليد الإيميل
+                    company_name = title or url.split("/")[2] if "/" in url else url
+                    data = generate_email(company_name, page_text[:600])
+                    if not data:
+                        continue
+
+                    ok, msg = send_email(target_email, data["subject"], data["body"])
+                    if ok:
+                        memory.mark(url)
+                        memory.mark_user(email_id)
+                        memory.log("Email", title, url, f"→ {target_email}", data["body"])
+                        antiban.record("email")
+                        print(f"✅ Email → {target_email} [{title[:30]}]")
+                        await asyncio.sleep(random.uniform(30, 60))
+
+                except Exception as e:
+                    print(f"  Email sub-error: {e}")
 
-                    try:
-                        await page.goto(url, timeout=20000)
-                        content = await page.content()
-                        emails = re.findall(r"[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}", content)
-                        valid = [e for e in emails if not e.endswith(('png','jpg','js','css','svg'))]
-                        
-                        if valid:
-                            target_email = valid[0]
-                            data = generate_cold_email_content(content, title)
-                            
-                            if data:
-                                ok, msg = send_smtp_email(target_email, data['subject'], data['body'])
-                                if ok:
-                                    LOCAL_STATS["email"] += 1
-                                    # 🛡️ تسجيل الرابط
-                                    log_manager.mark_processed(url)
-                                    log_manager.save_log_entry("Email", title, url, f"Target: {target_email}", data['body'])
-                                    print(f"✅ Email Sent: {title}")
-                                    await asyncio.sleep(20)
-                    except: pass
-            except: pass
+        except Exception as e:
+            print(f"Email search error: {e}")
+        finally:
             await browser.close()
-    except Exception as e:
-        print(f"Email Process Error: {e}")
 
-def run_email_loop_sync():
+def email_loop_sync():
     loop = asyncio.new_event_loop()
     asyncio.set_event_loop(loop)
     while True:
         try:
-            loop.run_until_complete(email_hunter_process())
-            time.sleep(300)
+            loop.run_until_complete(email_hunter())
         except Exception as e:
-            time.sleep(60)
-
-# --- Reddit Job ---
-def reddit_job():
-    try:
-        if LOCAL_STATS["reddit"] >= LIMITS["reddit_daily_comments"]: return
-        if not REDDIT_CLIENT_ID: return
-        
-        reddit = praw.Reddit(client_id=REDDIT_CLIENT_ID, client_secret=REDDIT_CLIENT_SECRET,
-                             username=REDDIT_USERNAME, password=REDDIT_PASSWORD, user_agent="NexusBot")
-        kw = random.choice(generate_polyglot_keywords())
-        for post in reddit.subreddit("all").search(kw, limit=3, sort="new"):
-            # 🛡️ الحماية الحديدية
-            if log_manager.is_processed(post.url): continue
-            
-            reply = generate_localized_reply(post.title + "\n" + post.selftext, "Reddit")
-            if reply:
-                post.reply(reply)
-                LOCAL_STATS["reddit"] += 1
-                # 🛡️ تسجيل الرابط
-                log_manager.mark_processed(post.url)
-                log_manager.save_log_entry("Reddit", post.title, post.url, post.selftext or "Image", reply)
-                print(f"✅ Reddit Replied: {post.title[:20]}")
-                break
-    except: pass
+            print(f"Email Loop Error: {e}")
+        time.sleep(random.randint(300, 600))
 
 # ==========================================
-# 6. التشغيل والواجهة
+# 13. تشغيل الخيوط
 # ==========================================
 
-def start_background_tasks():
-    threading.Thread(target=github_loop, daemon=True).start()
-    threading.Thread(target=run_email_loop_sync, daemon=True).start()
-    sched = BackgroundScheduler()
-    sched.add_job(reddit_job, 'interval', minutes=20)
-    sched.start()
-    print("🚀 All Nexus Systems Operational (With Iron Memory Protection)")
+def start_all():
+    threads = [
+        threading.Thread(target=_keyword_refresh_loop, daemon=True),   # مولّد الكلمات التلقائي
+        threading.Thread(target=github_loop, daemon=True),
+        threading.Thread(target=reddit_loop, daemon=True),
+        threading.Thread(target=hn_loop, daemon=True),
+        threading.Thread(target=devto_loop, daemon=True),
+        threading.Thread(target=discord_monitor_loop, daemon=True),
+        threading.Thread(target=email_loop_sync, daemon=True),
+    ]
+    for t in threads:
+        t.start()
 
-start_background_tasks()
+    print("🚀 Orgteh Marketing Engine — جميع الأنظمة تعمل")
 
-def get_dashboard_stats():
-    return LOCAL_STATS["reddit"], LOCAL_STATS["github"], LOCAL_STATS["email"]
+start_all()
 
-def get_chat_history():
-    try:
-        logs = log_manager.load_logs()
-        if not logs: logs = LOCAL_LOGS
-    except:
-        logs = LOCAL_LOGS
-        
-    formatted_chat = []
+# ==========================================
+# 14. واجهة Gradio
+# ==========================================
+
+def refresh_stats():
+    with _keyword_lock:
+        kw_count = len(_keyword_cache)
+    last_refresh = datetime.utcfromtimestamp(_last_keyword_refresh).strftime("%H:%M UTC") if _last_keyword_refresh else "لم يتم بعد"
+    return (
+        LOCAL_STATS.get("reddit", 0),
+        LOCAL_STATS.get("github", 0),
+        LOCAL_STATS.get("email", 0),
+        LOCAL_STATS.get("devto", 0),
+        LOCAL_STATS.get("hn", 0),
+        LOCAL_STATS.get("discord", 0),
+        f"{kw_count} كلمة | آخر تجديد: {last_refresh}",
+    )
+
+def refresh_logs():
+    logs = memory.load_logs()
+    chat = []
     for log in logs:
-        user_content = f"**[{log['platform']}] {log['title']}**\n\n{log.get('original_text','')}\n\n🔗 [Link]({log['url']})"
-        bot_content = f"**🗓️ {log['timestamp']}**\n\n{log['our_reply']}"
-        formatted_chat.append({"role": "user", "content": user_content})
-        formatted_chat.append({"role": "assistant", "content": bot_content})
-        
-    return formatted_chat
-
-with gr.Blocks(title="Nexus Ultimate") as demo:
-    gr.Markdown("# 🦅 Nexus Marketing Operations Center")
-    gr.Markdown(f"Storage: Upstash Redis | Engine: Playwright Docker | Contact: {TELEGRAM_CONTACT}")
-    
+        user_msg = (
+            f"**[{log['platform']}]** {log['title']}\n\n"
+            f"{log.get('snippet','')}\n\n"
+            f"🔗 {log['url']}"
+        )
+        bot_msg = f"🕐 {log['ts']}\n\n{log['reply']}"
+        chat.append({"role": "user", "content": user_msg})
+        chat.append({"role": "assistant", "content": bot_msg})
+    return chat
+
+with gr.Blocks(title="Orgteh Marketing Engine", theme=gr.themes.Base()) as demo:
+    gr.Markdown("""
+# 🚀 Orgteh Marketing Engine
+**موقع:** [orgteh.com](https://www.orgteh.com) | **المنصات:** GitHub · Reddit · HackerNews · Dev.to · Discord · Email
+""")
+
     with gr.Row():
-        r = gr.Number(label="Reddit Today")
-        g = gr.Number(label="GitHub Total")
-        e = gr.Number(label="Emails Today")
-    
-    refresh_btn = gr.Button("🔄 Refresh Data & Logs")
-
-    gr.Markdown("### 💬 Live Interaction Logs")
-    chatbot_view = gr.Chatbot(label="History Log", height=600)
-    
-    refresh_btn.click(get_dashboard_stats, outputs=[r, g, e])
-    refresh_btn.click(get_chat_history, outputs=[chatbot_view])
-    
-    demo.load(get_dashboard_stats, outputs=[r, g, e])
-    demo.load(get_chat_history, outputs=[chatbot_view])
+        r_reddit  = gr.Number(label="Reddit اليوم",    min_width=100)
+        r_github  = gr.Number(label="GitHub اليوم",    min_width=100)
+        r_email   = gr.Number(label="Emails اليوم",    min_width=100)
+        r_devto   = gr.Number(label="Dev.to اليوم",    min_width=100)
+        r_hn      = gr.Number(label="HN (للمراجعة)",   min_width=100)
+        r_discord = gr.Number(label="Discord DMs",     min_width=100)
+
+    r_keywords = gr.Textbox(label="🔑 كلمات البحث الحالية (AI مولّدة + ثابتة)", interactive=False)
+
+    refresh_btn = gr.Button("🔄 تحديث البيانات")
+    gr.Markdown("### 💬 سجل التفاعلات الحية")
+    log_view = gr.Chatbot(label="السجل", height=650, type="messages")
+
+    refresh_btn.click(refresh_stats, outputs=[r_reddit, r_github, r_email, r_devto, r_hn, r_discord, r_keywords])
+    refresh_btn.click(refresh_logs, outputs=[log_view])
+    demo.load(refresh_stats, outputs=[r_reddit, r_github, r_email, r_devto, r_hn, r_discord, r_keywords])
+    demo.load(refresh_logs, outputs=[log_view])
 
 if __name__ == "__main__":
-    demo.launch(server_name="0.0.0.0", server_port=7860, theme=gr.themes.Base())
+    demo.launch(server_name="0.0.0.0", server_port=7860)