feat: add Chainlit web UI + Dockerfile for Railway deployment

Adds a Chainlit-based web interface that wraps TradingAgentsGraph,
streaming analyst reports, research debates, and final decisions
to the browser in real-time. Configured for Anthropic Claude models.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

.chainlit/config.toml

@@ -0,0 +1,16 @@
+[project]
+enable_telemetry = false
+
+[UI]
+name = "TradingAgents"
+description = "Multi-Agent LLM Trading Analysis"
+default_collapse_content = true
+hide_cot = false
+
+[UI.theme.light]
+primary = "#1a73e8"
+background = "#ffffff"
+
+[UI.theme.dark]
+primary = "#00f0ff"
+background = "#09090b"

Dockerfile

@@ -0,0 +1,16 @@
+FROM python:3.13-slim
+
+WORKDIR /app
+
+# System deps for building Python packages
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    gcc g++ && \
+    rm -rf /var/lib/apt/lists/*
+
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+COPY . .
+
+# Chainlit listens on $PORT (Railway sets this automatically)
+CMD chainlit run app.py --host 0.0.0.0 --port ${PORT:-8000}

app.py

@@ -0,0 +1,183 @@
+"""Chainlit web UI for TradingAgents — deployed on Railway."""
+
+import os
+import re
+from datetime import date
+
+import chainlit as cl
+
+from tradingagents.graph.trading_graph import TradingAgentsGraph
+from tradingagents.default_config import DEFAULT_CONFIG
+from cli.stats_handler import StatsCallbackHandler
+
+
+def parse_ticker_date(text: str):
+    """Extract ticker symbol and optional date from user message.
+
+    Examples:
+        "NVDA"                       -> ("NVDA", today)
+        "Analyze AAPL 2024-12-01"    -> ("AAPL", "2024-12-01")
+        "What about TSLA?"           -> ("TSLA", today)
+    """
+    # Try to find a date (YYYY-MM-DD)
+    date_match = re.search(r"(\d{4}-\d{2}-\d{2})", text)
+    trade_date = date_match.group(1) if date_match else str(date.today())
+
+    # Find uppercase 1-5 letter words as candidate tickers
+    candidates = re.findall(r"\b([A-Z]{1,5})\b", text)
+    # Filter out common English words
+    skip = {"I", "A", "THE", "AND", "OR", "FOR", "TO", "IN", "ON", "AT", "IS",
+            "IT", "OF", "BY", "AS", "AN", "BE", "IF", "SO", "DO", "MY", "UP",
+            "NO", "NOT", "ALL", "BUT", "HOW", "GET", "HAS", "HAD", "CAN",
+            "WHAT", "ABOUT", "BUY", "SELL", "HOLD"}
+    tickers = [c for c in candidates if c not in skip]
+
+    ticker = tickers[0] if tickers else None
+    return ticker, trade_date
+
+
+def build_config():
+    """Build TradingAgents config for Anthropic/Claude."""
+    config = DEFAULT_CONFIG.copy()
+    config["llm_provider"] = "anthropic"
+    config["deep_think_llm"] = os.getenv("DEEP_THINK_MODEL", "claude-sonnet-4-5-20241022")
+    config["quick_think_llm"] = os.getenv("QUICK_THINK_MODEL", "claude-haiku-4-5-20251001")
+    config["backend_url"] = None
+    config["max_debate_rounds"] = 1
+    config["max_risk_discuss_rounds"] = 1
+    config["data_vendors"] = {
+        "core_stock_apis": "yfinance",
+        "technical_indicators": "yfinance",
+        "fundamental_data": "yfinance",
+        "news_data": "yfinance",
+    }
+    return config
+
+
+# Report field -> display name
+REPORT_NAMES = {
+    "market_report": "Market Analyst",
+    "sentiment_report": "Sentiment Analyst",
+    "news_report": "News Analyst",
+    "fundamentals_report": "Fundamentals Analyst",
+}
+
+
+@cl.on_chat_start
+async def on_chat_start():
+    await cl.Message(
+        content=(
+            "**TradingAgents** — Multi-Agent LLM Trading Analysis\n\n"
+            "Send a ticker symbol to analyze. Examples:\n"
+            "- `NVDA`\n"
+            "- `Analyze AAPL 2024-12-01`\n"
+            "- `What's the outlook for TSLA?`\n\n"
+            "I'll run a team of AI analysts, researchers, traders, and risk managers "
+            "to produce a trading decision."
+        )
+    ).send()
+
+
+@cl.on_message
+async def on_message(message: cl.Message):
+    ticker, trade_date = parse_ticker_date(message.content)
+
+    if not ticker:
+        await cl.Message(
+            content="I couldn't find a ticker symbol. Try something like `NVDA` or `Analyze AAPL 2024-12-01`."
+        ).send()
+        return
+
+    # Status message
+    status_msg = cl.Message(content=f"Analyzing **{ticker}** for **{trade_date}**...")
+    await status_msg.send()
+
+    # Build graph
+    config = build_config()
+    stats = StatsCallbackHandler()
+
+    try:
+        graph = TradingAgentsGraph(
+            selected_analysts=["market", "social", "news", "fundamentals"],
+            debug=False,
+            config=config,
+            callbacks=[stats],
+        )
+    except Exception as e:
+        await cl.Message(content=f"Failed to initialize agents: {e}").send()
+        return
+
+    # Create initial state and stream
+    init_state = graph.propagator.create_initial_state(ticker, trade_date)
+    args = graph.propagator.get_graph_args(callbacks=[stats])
+
+    # Track which reports/phases we've already shown
+    seen_reports = set()
+    seen_debate = False
+    seen_risk = False
+    seen_trader = False
+    final_state = None
+
+    try:
+        async for chunk in graph.graph.astream(init_state, **args):
+            final_state = chunk
+
+            # --- Analyst reports ---
+            for field, name in REPORT_NAMES.items():
+                if field not in seen_reports and chunk.get(field):
+                    seen_reports.add(field)
+                    report = chunk[field]
+                    # Show as a collapsible Step
+                    async with cl.Step(name=f"{name} Report", type="tool") as step:
+                        step.output = report[:3000] if len(report) > 3000 else report
+
+            # --- Investment debate (Bull vs Bear) ---
+            debate = chunk.get("investment_debate_state")
+            if debate and not seen_debate and debate.get("judge_decision"):
+                seen_debate = True
+                async with cl.Step(name="Research Debate", type="tool") as step:
+                    step.output = (
+                        f"**Judge Decision:**\n{debate['judge_decision']}"
+                    )
+
+            # --- Trader plan ---
+            if not seen_trader and chunk.get("trader_investment_plan"):
+                seen_trader = True
+                async with cl.Step(name="Trader Plan", type="tool") as step:
+                    step.output = chunk["trader_investment_plan"][:3000]
+
+            # --- Risk debate ---
+            risk = chunk.get("risk_debate_state")
+            if risk and not seen_risk and risk.get("judge_decision"):
+                seen_risk = True
+                async with cl.Step(name="Risk Assessment", type="tool") as step:
+                    step.output = f"**Risk Decision:**\n{risk['judge_decision']}"
+
+    except Exception as e:
+        await cl.Message(content=f"Error during analysis: {e}").send()
+        return
+
+    if not final_state:
+        await cl.Message(content="Analysis produced no results.").send()
+        return
+
+    # Process final decision
+    decision_text = final_state.get("final_trade_decision", "No decision reached.")
+    signal = graph.process_signal(decision_text)
+
+    # Stats summary
+    s = stats.get_stats()
+    stats_line = (
+        f"*{s['llm_calls']} LLM calls · {s['tool_calls']} tool calls · "
+        f"{s['tokens_in']:,} tokens in · {s['tokens_out']:,} tokens out*"
+    )
+
+    await cl.Message(
+        content=(
+            f"## {ticker} — Trading Decision\n\n"
+            f"**Signal: {signal}**\n\n"
+            f"---\n\n"
+            f"{decision_text}\n\n"
+            f"---\n{stats_line}"
+        )
+    ).send()

railway.toml

@@ -0,0 +1,8 @@
+[build]
+builder = "DOCKERFILE"
+dockerfilePath = "Dockerfile"
+
+[deploy]
+healthcheckPath = "/"
+restartPolicyType = "ON_FAILURE"
+restartPolicyMaxRetries = 3