feat: add complete multi-agent AI system with Telegram bot

Initial implementation of MultiAgentsAI featuring:
- Dynamic agent configuration via environment variables
- SQLite database and JSON daily context storage
- OpenRouter API integration for multiple AI models
- Telegram bot with commands for control, history, and interaction
- Autonomous conversation loop with pause/resume and agent silencing
- Model testing and statistics commands

context_manager.py

@@ -0,0 +1,51 @@
+# Gerado por IA
+import os
+from datetime import datetime
+from database import get_today_messages, save_message
+from config import settings
+
+
+def get_daily_context_path() -> str:
+    today = datetime.utcnow().strftime("%Y-%m-%d")
+    os.makedirs(settings.CONTEXTS_DIR, exist_ok=True)
+    return os.path.join(settings.CONTEXTS_DIR, f"conversas_{today}.json")
+
+
+def load_daily_context() -> list[dict]:
+    """Carrega o contexto do dia. SQLite e a fonte da verdade; JSON e um cache sincronizado."""
+    # Fonte primaria: SQLite
+    context = get_today_messages()
+    if context:
+        return context
+    # Fallback: JSON (caso SQLite esteja vazio)
+    path = get_daily_context_path()
+    if os.path.exists(path):
+        try:
+            import json
+            with open(path, "r", encoding="utf-8") as f:
+                return json.load(f)
+        except Exception:
+            return []
+    return []
+
+
+def append_to_context(agent_name: str, message: str) -> None:
+    path = get_daily_context_path()
+    context = load_daily_context()
+    timestamp = datetime.utcnow().isoformat()
+    entry = {"agent_name": agent_name, "message": message, "created_at": timestamp}
+    context.append(entry)
+
+    import json
+    with open(path, "w", encoding="utf-8") as f:
+        json.dump(context, f, ensure_ascii=False, indent=2)
+
+    save_message(agent_name, message)
+
+
+def get_context_summary(max_messages: int = 50) -> str:
+    context = load_daily_context()
+    if not context:
+        return "Nenhuma conversa registrada para hoje ainda."
+    lines = [f"[{c['created_at']}] {c['agent_name']}: {c['message']}" for c in context[-max_messages:]]
+    return "\n".join(lines)