Claude File Management Server

# import asyncio # import nest_asyncio # nest_asyncio.apply() # from llama_index.llms.ollama import Ollama # from llama_index.core import Settings # from llama_index.tools.mcp import BasicMCPClient, McpToolSpec # from llama_index.core.agent.workflow import FunctionAgent, ToolCall, ToolCallResult # from llama_index.core.workflow import Context # # Initialize LLM # # llm = Ollama(model="mistral") # llm = Ollama(model="mistral") # Settings.llm = llm # # System prompt for the agent # SYSTEM_PROMPT = """\ # You are an AI assistant for Tool Calling. # Before you help a user, you need to work with tools to interact with Our Database # """ # # Create agent # async def get_agent(tools: McpToolSpec): # tools = await tools.to_tool_list_async() # agent = FunctionAgent( # name="Agent", # description="An agent that can work with Our Database software.", # tools=tools, # llm=llm, # system_prompt=SYSTEM_PROMPT, # ) # return agent # # Handle user messages # async def handle_user_message( # message_content: str, # agent: FunctionAgent, # agent_context: Context, # verbose: bool = False, # ): # handler = agent.run(message_content, ctx=agent_context) # async for event in handler.stream_events(): # if verbose and isinstance(event, ToolCall): # print(f"Calling tool {event.tool_name} with kwargs {event.tool_kwargs}") # elif verbose and isinstance(event, ToolCallResult): # print(f"Tool {event.tool_name} returned {event.tool_output}") # response = await handler # return str(response) # # Main entry point # async def main(): # mcp_client = BasicMCPClient("http://127.0.0.1:8000/sse") # mcp_tool = McpToolSpec(client=mcp_client) # # Initialize agent # agent = await get_agent(mcp_tool) # agent_context = Context(agent) # # REPL loop # while True: # user_input = input("Enter your message: ") # if user_input.strip().lower() == "exit": # print("Exiting...") # break # print("User: ", user_input) # response = await handle_user_message(user_input, agent, agent_context, verbose=True) # print("Agent: ", response) # if __name__ == "__main__": # asyncio.run(main()) # import asyncio # from mcp.client.session import ClientSession # from mcp.types import CallToolRequest # from mcp.client.stdio import connect_stdio # ✅ correct function # import ollama # import textwrap # MODEL = "llama3.2:3b" # CHUNK_SIZE = 20000 # def chunk_text(text: str, size: int = CHUNK_SIZE): # return textwrap.wrap(text, size) # def summarize_with_ollama(prompt: str) -> str: # response = ollama.chat( # model=MODEL, # messages=[{"role": "user", "content": prompt}] # ) # return response["message"]["content"].strip() # def summarize_pdf_text(text: str) -> str: # chunks = chunk_text(text, CHUNK_SIZE) # partial_summaries = [] # print(f"📄 Splitting PDF into {len(chunks)} chunks...") # for i, chunk in enumerate(chunks, 1): # print(f"📝 Summarizing chunk {i}/{len(chunks)}...") # prompt = f"Summarize the following PDF content into bullet points:\n\n{chunk}" # summary = summarize_with_ollama(prompt) # partial_summaries.append(summary) # print("🔄 Generating final consolidated summary...") # merged_prompt = "Combine these chunk summaries into one coherent executive summary:\n\n" + "\n\n".join(partial_summaries) # return summarize_with_ollama(merged_prompt) # async def main(): # print("🚀 Starting PDF Summarizer Agent with ollama.chat...") # async with connect_stdio("server.py") as (read_stream, write_stream): # client = ClientSession(read_stream, write_stream) # await client.initialize() # print("✅ Connected to MCP server!") # while True: # user_input = input("\n👤 Enter command (e.g. summarize pdf file.pdf or exit): ") # if user_input.strip().lower() == "exit": # break # if user_input.startswith("summarize pdf"): # _, _, file_path = user_input.partition("pdf") # file_path = file_path.strip() # tool_call = CallToolRequest(name="read_pdf", arguments={"file_path": file_path}) # result = await client.call_tool(tool_call) # if result.content and hasattr(result.content[0], "text"): # pdf_text = result.content[0].text # print("📄 Extracted text, sending to Ollama...") # summary = summarize_pdf_text(pdf_text) # print("\n📑 Final Summary:\n", summary) # else: # print("❌ Could not extract text") # if __name__ == "__main__": # asyncio.run(main()) # import asyncio # import nest_asyncio # import httpx # nest_asyncio.apply() # from llama_index.llms.ollama import Ollama # from llama_index.core import Settings # from llama_index.tools.mcp import BasicMCPClient, McpToolSpec # from llama_index.core.agent.workflow import FunctionAgent, ToolCall, ToolCallResult # from llama_index.core.workflow import Context # # ========================================================== # # CONFIG # # ========================================================== # # 🔗 Replace with your ngrok public FastAPI endpoint # FASTAPI_URL = "https://37b11cbf9e1c.ngrok-free.app/generate" # # Initialize LLM # llm = Ollama( # model="mistral", # temperature=0.0, # Deterministic # top_p=0.9, # repeat_penalty=1.1, # ) # Settings.llm = llm # SYSTEM_PROMPT = """\ # You are a helpful AI assistant that MUST use tools to complete tasks. # RULES: # 1. ALWAYS use tools when asked about data, files, or databases # 2. Do NOT make up information - use tools to get real data # 3. For database tasks: use database tools (create_table, insert_data, select_data) # 4. For file tasks: use file tools (list_files, read_json_file, write_json_file) # 5. Break complex tasks into simple steps # 6. Use one tool at a time and explain what you're doing # """ # # ========================================================== # # MCP Agent Setup # # ========================================================== # async def get_agent(tools: McpToolSpec): # """Create and configure the function agent with MCP tools""" # tools_list = await tools.to_tool_list_async() # print(f"✅ Loaded {len(tools_list)} tools from MCP server") # agent = FunctionAgent( # name="MCPAgent", # description="An agent that can work with databases and files through MCP tools.", # tools=tools_list, # llm=llm, # system_prompt=SYSTEM_PROMPT, # max_function_calls=5, # verbose=True, # ) # return agent # async def handle_user_message(message_content: str, agent: FunctionAgent, agent_context: Context): # """Handle user messages with tool usage""" # handler = agent.run(message_content, ctx=agent_context) # tool_output_accum = [] # collect tool outputs # async for event in handler.stream_events(): # if isinstance(event, ToolCall): # print(f"🔧 Calling tool: {event.tool_name}") # print(f" Parameters: {event.tool_kwargs}") # elif isinstance(event, ToolCallResult): # print(f"✅ Tool {event.tool_name} completed") # print(f" Result: {str(event.tool_output)[:200]}...") # tool_output_accum.append(str(event.tool_output)) # response = await handler # return "\n".join(tool_output_accum), str(response) # # ========================================================== # # FastAPI Communication # # ========================================================== # async def send_to_fastapi(data: str, query: str): # """Send payload to FastAPI server (Colab via ngrok)""" # payload = {"data": data, "query": query} # async with httpx.AsyncClient(timeout=120.0, verify=False) as client: # resp = await client.post(FASTAPI_URL, json=payload) # resp.raise_for_status() # return resp.json() # # ========================================================== # # Main # # ========================================================== # async def main(): # print("🚀 Starting MCP Agent...") # mcp_client = BasicMCPClient("http://127.0.0.1:8000/sse") # mcp_tool = McpToolSpec(client=mcp_client) # # Initialize agent # agent = await get_agent(mcp_tool) # agent_context = Context(agent) # print("✅ Agent ready! Type 'exit' to quit.") # print("-" * 50) # while True: # user_input = input("\n👤 Enter your message: ").strip() # if user_input.lower() == "exit": # print("👋 Exiting...") # break # print(f"\n👤 User: {user_input}") # print("🤖 Agent: Processing...") # tool_result, agent_response = await handle_user_message(user_input, agent, agent_context) # # Send tool outputs + user query to FastAPI # try: # llm_response = await send_to_fastapi(tool_result, user_input) # print(f"\n🤖 Final Answer (from FastAPI LLM): {llm_response}") # except Exception as e: # print(f"❌ Error sending to FastAPI: {e}") # if __name__ == "__main__": # asyncio.run(main())