feat: implement core tool infrastructure, Redis integration, and comprehensive test suite for agent functionality

config.py

@@ -3,7 +3,28 @@ from dotenv import load_dotenv
 
 load_dotenv()
 
-ANTHROPIC_API_KEY = os.getenv("ANTHROPIC_API_KEY", "")
-OPENAI_API_KEY = os.getenv("OPENAI_API_KEY", "")
-DEFAULT_MODEL = os.getenv("DEFAULT_MODEL", "claude-sonnet-4-20250514")
+# Common
 LOG_LEVEL = os.getenv("LOG_LEVEL", "INFO")
+
+# Anthropic (Legacy)
+ANTHROPIC_API_KEY = os.getenv("ANTHROPIC_API_KEY", "")
+
+# Google Gemini
+GEMINI_API_KEY = os.getenv("GEMINI_API_KEY", "")
+DEFAULT_MODEL = os.getenv("DEFAULT_MODEL", "gemini-1.5-flash")
+
+# Qwen API
+QWEN_API_KEY = os.getenv("QWEN_API_KEY", "")
+QWEN_BASE_URL = os.getenv("QWEN_BASE_URL", "https://dashscope.aliyuncs.com/compatible-mode/v1")
+QWEN_MODEL = os.getenv("QWEN_MODEL", "qwen-plus")
+
+# Local LLM config
+USE_LOCAL_LLM = os.getenv("USE_LOCAL_LLM", "false").lower() == "true"
+LOCAL_MODEL_ID = os.getenv("LOCAL_MODEL_ID", "Qwen/Qwen2.5-0.5B-Instruct")
+
+# Redis
+REDIS_HOST = os.getenv("REDIS_HOST", "localhost")
+REDIS_PORT = int(os.getenv("REDIS_PORT", 6379))
+REDIS_DB = int(os.getenv("REDIS_DB", 0))
+REDIS_PASSWORD = os.getenv("REDIS_PASSWORD", None)
+REDIS_KEY_PREFIX = os.getenv("REDIS_KEY_PREFIX", "chat")

llm.py

@@ -0,0 +1,17 @@
+from langchain_google_genai import ChatGoogleGenerativeAI
+from src.config import GEMINI_API_KEY, DEFAULT_MODEL
+
+llm = ChatGoogleGenerativeAI(
+    model=DEFAULT_MODEL,
+    temperature=0,
+    top_p=1,
+    top_k=1,
+    max_tokens=None,
+    timeout=None,
+    max_retries=2,
+    google_api_key=GEMINI_API_KEY
+)
+
+if __name__ == "__main__":
+    response = llm.invoke("Hello World là gì?").content
+    print(response)

tools/base.py

@@ -9,10 +9,11 @@ from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
 from langchain_core.tools import Tool
 from langchain_openai import ChatOpenAI
 from langchain_huggingface import HuggingFacePipeline
+from langchain_google_genai import ChatGoogleGenerativeAI
 try:
-    from ..config import QWEN_API_KEY, QWEN_BASE_URL, QWEN_MODEL, LOG_LEVEL, USE_LOCAL_LLM, LOCAL_MODEL_ID
+    from ..config import QWEN_API_KEY, QWEN_BASE_URL, QWEN_MODEL, LOG_LEVEL, USE_LOCAL_LLM, LOCAL_MODEL_ID, GEMINI_API_KEY
 except (ImportError, ValueError):
-    from config import QWEN_API_KEY, QWEN_BASE_URL, QWEN_MODEL, LOG_LEVEL, USE_LOCAL_LLM, LOCAL_MODEL_ID
+    from config import QWEN_API_KEY, QWEN_BASE_URL, QWEN_MODEL, LOG_LEVEL, USE_LOCAL_LLM, LOCAL_MODEL_ID, GEMINI_API_KEY
 
 logger = logging.getLogger(__name__)
 
@@ -35,8 +36,18 @@ def register_tool(name: str, description: str, parameters: list):
 
 def get_llm():
     """
-    Initialize and return the LLM based on configuration (Local vs Cloud).
+    Initialize and return the LLM based on configuration (Gemini > Local > Cloud Qwen).
     """
+    # 1. Prioritize Gemini if API key is present
+    if GEMINI_API_KEY and not GEMINI_API_KEY.startswith("your-"):
+        logger.info("Initializing Google Gemini LLM...")
+        return ChatGoogleGenerativeAI(
+            model="gemini-2.0-flash", # or from config
+            google_api_key=GEMINI_API_KEY,
+            temperature=0.1,
+        )
+
+    # 2. Local LLM
     if USE_LOCAL_LLM:
         logger.info(f"Loading Local LLM: {LOCAL_MODEL_ID}...")
         try:

tools/summarizer.py

@@ -11,6 +11,10 @@ from langchain_core.prompts import ChatPromptTemplate
 
 from .base import register_tool, get_llm
 from .utils import preprocess_messages, extract_metadata_from_messages
+try:
+    from ..redis_client import redis_client
+except (ImportError, ValueError):
+    from redis_client import redis_client
 
 # --- Pydantic Schemas ---
 class ThreadSummary(BaseModel):

tools/utils.py

@@ -1,5 +1,6 @@
 import re
 import json
+from collections import defaultdict
 
 def extract_metadata_from_messages(messages: list[dict]) -> dict:
     """