openkite 0.1.2

OpenKite trading assistant packaged as a modular Python application.

---

Overview

OpenKite is an AI-powered agentic trading assistant that combines real-time market access, fundamental analysis, and web intelligence into a single intelligent workflow. Built on the OpenAI Agents SDK, it autonomously selects and orchestrates tools to answer complex trading and investment queries — without you having to switch between multiple platforms.

Features

	Feature	Description
🤖	Agentic Reasoning	Dynamically decides which tools to invoke based on user intent; minimizes unnecessary calls while maximizing insight
📈	Zerodha Kite Integration	Fetch portfolio, positions, and holdings; place orders (including GTT); access real-time market data via the official Kite MCP server
🌐	Web Intelligence via Firecrawl	Pull the latest news, trends, and external signals to enhance decision-making with real-world context
📊	Fundamental Analysis via Screener.in	Retrieve company financials, key ratios, and quarterly trends for long-term investment research via a custom-built CLI tool — screenercli

Architecture

The agent orchestrator sits between the OpenAI Agents SDK and three tool groups:

• Remote MCP — Zerodha Kite MCP server for live trading operations
• Local MCP — Firecrawl MCP server for web crawl and search
• Custom CLI Tools — Screener.in CLI for fundamental data lookups

Installation

pip install openkite

For local development:

pip install -e .[dev]

Prerequisite: Node.js (npx) is required to run the Firecrawl MCP server.

Configuration

Create a .env file in your working directory (or supply it via --env-file):

OPENAI_API_KEY=sk-...
FIRECRAWL_API_KEY=fc-...
OPENKITE_MODEL=gpt-4o
ZERODHA_MCP_SSE_URL=https://mcp.kite.trade/sse
OPENKITE_DB_PATH=~/.openkite/chat_history.db
OPENKITE_HOST=127.0.0.1
OPENKITE_PORT=7860
OPENKITE_SHARE=false

Keep .env out of source control — add it to .gitignore.

Variable	Description
OPENAI_API_KEY	Your OpenAI API key
FIRECRAWL_API_KEY	Your Firecrawl API key
OPENKITE_MODEL	OpenAI model to use (e.g. gpt-4o)
ZERODHA_MCP_SSE_URL	Zerodha Kite MCP SSE endpoint
OPENKITE_DB_PATH	Path for persisting chat history
OPENKITE_HOST	Host for the Gradio UI server
OPENKITE_PORT	Port for the Gradio UI server
OPENKITE_SHARE	Set true to create a public Gradio share link

Usage

Launch the chat UI:

openkite chat

Validate your configuration:

openkite check

Pass custom runtime options:

openkite chat --env-file .env --host 0.0.0.0 --port 7860

Example Prompts

Try these in the chat UI to exercise Kite MCP, Firecrawl, and Screener together:

Portfolio & Positions

• Show my current holdings with live prices, day change, unrealised P&L, and portfolio weights.
• List my open positions and pending orders, then flag anything that needs attention today.
• Review my top holdings by value using live Kite data, Screener fundamentals, and recent news, then highlight key risks and opportunities without placing orders.

Fundamental Analysis

• Compare TCS and Infosys : valuation, growth, margins, ROCE, and the latest quarterly trend.
• Show the last 4 quarters for HDFCBANK and tell me whether revenue, profit, and return ratios are improving or weakening.

Market Intelligence

• Use recent web coverage to summarize the most important Reliance Industries news from the last 7 days and explain the likely market impact.
• For NSE:INFY, combine live Kite price action, fundamentals, and recent web news into a short trading and investment brief.

Screenshots

Chat interface with example prompt suggestions

AI-generated portfolio analysis with sell recommendations

Live agent tool-call trace during a Nifty 50 scan