QuantTradingOS Chatbot

Status: Active · Layer: Core · Integration: OS-integrated

Chat UI for QuantTradingOS: ask in natural language for backtests, prices, news, insider data, or to run the pipeline. Uses LangChain + MCP tools (the mcp-server connects to the orchestrator and data-ingestion-service).

Prerequisites

• Orchestrator and data-ingestion-service running (e.g. docker-compose -f orchestrator/docker-compose.full.yml up).
• mcp-server as a sibling directory of chatbot/ (or set MCP_SERVER_CWD).
• OpenAI API key (for the LLM).

Setup

cd chatbot
pip install -r requirements.txt
cp config/env.example config/.env
# Edit config/.env: set OPENAI_API_KEY=

Run

streamlit run app.py

Open the URL shown (default http://localhost:8501). Ask e.g.:

• "Run a backtest on AAPL for the last year"
• "Get the latest prices for SPY"
• "What's the recent news for Tesla?"
• "Run the pipeline" (run_decision)

Environment

Variable	Default	Description
OPENAI_API_KEY	—	Required. OpenAI API key for the chat model.
MCP_SERVER_COMMAND	python	Executable to run the MCP server (use python3.11 if needed).
MCP_SERVER_ARGS	-m,qtos_mcp.server	Comma-separated args to start the MCP server.
MCP_SERVER_CWD	../mcp-server	Working directory for the MCP server process.

How it works

1. Streamlit provides the chat UI.
2. LangChain builds an agent with tools loaded from the QuantTradingOS MCP server (via stdio).
3. The MCP server is spawned by the chatbot and translates tool calls into REST requests to the orchestrator and data service.
4. The LLM (OpenAI) decides when to call which tool and formats the reply.

License

MIT (same as QuantTradingOS).