Modern financial analysis is rapidly moving toward automation and agentic workflows. Integrating large language models (LLMs) with real-time financial data unlocks not just powerful insights but also entirely new ways of interacting with portfolio data.<\/p>\n\n\n\n
This post walks through a practical, autonomous solution using Google ADK<\/a>, Zerodha\u2019s Kite MCP<\/a> protocol, and an LLM for actionable portfolio analytics. The full workflow and code are available on GitHub<\/a>.<\/p>\n\n\n\n The workflow has three main steps:<\/p>\n\n\n\n All API keys and connection details are managed through environment variables for security and reproducibility.<\/p>\n\n\n\n All API keys and MCP endpoints are managed via environment variables or a Example
\n\n\n\nWhy This Stack?<\/h2>\n\n\n\n
\n\n\n\nArchitecture Overview<\/h2>\n\n\n\n
get_holdings<\/code> tool.<\/li>
\n\n\n\nKey Code Snippets<\/h2>\n\n\n\n
1. Environment and Dependency Setup<\/h3>\n\n\n\n
import os\nfrom dotenv import load_dotenv\n\n# Load API keys and config from .env\nload_dotenv('.env')\nos.environ[\"GOOGLE_API_KEY\"] = os.environ[\"GOOGLE_API_KEY\"]\nos.environ[\"GOOGLE_GENAI_USE_VERTEXAI\"] = \"False\"\n<\/code><\/pre>\n\n\n\n
\n\n\n\n2. ADK Agent and Toolset Initialization<\/h3>\n\n\n\n
from google.adk.agents.llm_agent import LlmAgent\nfrom google.adk.tools.mcp_tool.mcp_toolset import MCPToolset, SseServerParams\n\nMCP_SSE_URL = os.environ.get(\"MCP_SSE_URL\", \"https:\/\/mcp.kite.trade\/sse\")\n\ntoolset = MCPToolset(\n connection_params=SseServerParams(url=MCP_SSE_URL, headers={})\n)\n\nroot_agent = LlmAgent(\n model='gemini-2.0-flash',\n name='zerodha_portfolio_assistant',\n instruction=(\n \"You are an expert Zerodha portfolio assistant. \"\n \"Use the 'login' tool to authenticate, and the 'get_holdings' tool to fetch stock holdings. \"\n \"When given portfolio data, analyze for concentration risk and best\/worst performers.\"\n ),\n tools=[toolset]\n)\n<\/code><\/pre>\n\n\n\n
\n\n\n\n3. Orchestrating the Workflow<\/h3>\n\n\n\n
from google.adk.sessions import InMemorySessionService\nfrom google.adk.artifacts.in_memory_artifact_service import InMemoryArtifactService\nfrom google.adk.runners import Runner\nfrom google.genai import types\n\nimport asyncio\n\nasync def run_workflow():\n session_service = InMemorySessionService()\n artifacts_service = InMemoryArtifactService()\n session = await session_service.create_session(\n state={}, app_name='zerodha_portfolio_app', user_id='user1'\n )\n\n runner = Runner(\n app_name='zerodha_portfolio_app',\n agent=root_agent,\n artifact_service=artifacts_service,\n session_service=session_service,\n )\n\n # 1. Login Step\n login_query = \"Authenticate and provide the login URL for Zerodha.\"\n content = types.Content(role='user', parts=[types.Part(text=login_query)])\n login_url = None\n async for event in runner.run_async(session_id=session.id, user_id=session.user_id, new_message=content):\n if event.is_final_response():\n import re\n match = re.search(r'(https?:\/\/[^\\s)]+)', getattr(event.content.parts[0], \"text\", \"\"))\n if match:\n login_url = match.group(1)\n if not login_url:\n print(\"No login URL found. Exiting.\")\n return\n print(f\"Open this URL in your browser to authenticate:\\n{login_url}\")\n import webbrowser; webbrowser.open(login_url)\n input(\"Press Enter after completing login...\")\n\n # 2. Fetch Holdings\n holdings_query = \"Show my current stock holdings.\"\n content = types.Content(role='user', parts=[types.Part(text=holdings_query)])\n holdings_raw = None\n async for event in runner.run_async(session_id=session.id, user_id=session.user_id, new_message=content):\n if event.is_final_response():\n holdings_raw = getattr(event.content.parts[0], \"text\", None)\n if not holdings_raw:\n print(\"No holdings data found.\")\n return\n\n # 3. Analysis\n analysis_prompt = f\"\"\"\nYou are a senior portfolio analyst.\n\nGiven only the raw stock holdings listed below, do not invent or assume any other holdings.\n\n1. **Concentration Risk**: Identify if a significant percentage of the total portfolio is allocated to a single stock or sector. Quantify the largest exposures, explain why this matters, and suggest specific diversification improvements.\n\n2. **Performance Standouts**: Clearly identify the best and worst performing stocks in the portfolio (by absolute and percentage P&L), and give actionable recommendations.\n\nRaw holdings:\n\n{holdings_raw}\n\nUse only the provided data.\n\"\"\"\n content = types.Content(role='user', parts=[types.Part(text=analysis_prompt)])\n async for event in runner.run_async(session_id=session.id, user_id=session.user_id, new_message=content):\n if event.is_final_response():\n print(\"\\n=== Portfolio Analysis Report ===\\n\")\n print(getattr(event.content.parts[0], \"text\", \"\"))\n\nasyncio.run(run_workflow())\n<\/code><\/pre>\n\n\n\n
\n\n\n\nSecurity and Environment Configuration<\/h2>\n\n\n\n
.env<\/code> file.
Never hardcode sensitive information in code.<\/p>\n\n\n\n.env<\/code> file:<\/strong><\/p>\n\n\n\nGOOGLE_API_KEY=your_google_gemini_api_key\nMCP_SSE_URL=https:\/\/mcp.kite.trade\/sse\n<\/code><\/pre>\n\n\n\n
\n\n\n\nWhat This Enables<\/h2>\n\n\n\n