dashboard

name	dashboard
description	Scaffold a Streamlit-based financial dashboard using Massive APIs. Use when building market data visualizations, multi-asset dashboards, or real-time monitoring interfaces.
argument-hint	[project-name] [focus: multi-asset|options|crypto|macro]
disable-model-invocation	false
allowed-tools	Write Edit Bash Read

Scaffold a Streamlit financial dashboard

Project name: $0 (default: dashboard if not specified) Focus area: $1 (default: multi-asset if not specified)

Keep the app minimal. One file, direct RESTClient() calls wrapped in @st.cache_data, straightforward Plotly charts. Do not add env-var configuration layers, plugin systems, or utility modules (require_env, value_from, first_available, format helpers) beyond what Streamlit natively needs. A working 80-line streamlit_app.py beats a 300-line multi-module dashboard.

Brand rules (apply to ALL generated files)

• No emojis anywhere, including Streamlit page_icon. Use a text string or None instead.
• No em dashes. Use commas, periods, semicolons, or parentheses. For NA/missing values in tables, use "N/A" or "-", not an em dash.
• The only valid API domain is api.massive.com. Never use any other API domain in generated code.

Architecture

Default layout — single file:

$0/
  streamlit_app.py       # Everything: client, caching, sidebar, charts
  pyproject.toml
  .env.example
  README.md

Start here. Put @st.cache_resource client, @st.cache_data API wrappers, sidebar controls, and Plotly charts all in streamlit_app.py. Expect ~80-150 lines for most focus areas.

Multi-module layout — only when the user explicitly asks for a "terminal," "Bloomberg-style," or "multi-panel" app. In that case (and only that case), split into:

$0/
  streamlit_app.py       # Entry point: layout, styling, sidebar
  terminal/
    data.py              # API calls + caching
    config.py            # Colors, TTL constants
    charts.py            # Chart rendering functions
    panels/              # One file per panel
  .streamlit/
    config.toml          # Theme configuration
  pyproject.toml
  .env.example
  README.md

Do not introduce the multi-module layout for a plain "dashboard" or any focus area the user didn't describe as terminal-grade.

Key patterns

Client singleton (never recreate per call)

import streamlit as st

@st.cache_resource
def _get_client(api_key: str):
    from massive import RESTClient
    return RESTClient(api_key=api_key)

TTL-based caching (reduce API calls)

TTL_SNAPSHOT = 30     # live quotes refresh every 30s
TTL_CHART = 60        # historical bars refresh every 60s
TTL_OPTIONS = 45      # options chain refresh every 45s
TTL_NEWS = 120        # news refresh every 2 min
TTL_MACRO = 3600      # macro data refresh every hour

@st.cache_data(ttl=TTL_SNAPSHOT, show_spinner=False)
def get_snapshot(api_key: str, tickers: tuple[str, ...]) -> dict:
    client = _get_client(api_key)
    return {s.ticker: s for s in client.list_universal_snapshots(ticker_any_of=list(tickers))}

@st.cache_data(ttl=TTL_CHART, show_spinner=False)
def get_aggs(api_key: str, ticker: str, multiplier: int, timespan: str, lookback_days: int) -> list:
    client = _get_client(api_key)
    from datetime import datetime, timedelta
    from itertools import islice
    to_date = datetime.now().strftime("%Y-%m-%d")
    from_date = (datetime.now() - timedelta(days=lookback_days)).strftime("%Y-%m-%d")
    return list(islice(client.list_aggs(ticker, multiplier, timespan, from_date, to_date, sort="asc"), 5000))

Always pass sort="asc" to list_aggs() for chronological data.

Note: pass api_key as a plain string and ticker lists as tuples so they are hashable for st.cache_data.

Plotly dark theme charts

import plotly.graph_objects as go

fig = go.Figure(data=[go.Candlestick(
    x=dates, open=opens, high=highs, low=lows, close=closes
)])
fig.update_layout(
    template="plotly_dark",
    paper_bgcolor="#000000",
    plot_bgcolor="#0a0a0a",
    xaxis_rangeslider_visible=False,
)

Technical indicators (SMA, EMA, RSI, MACD)

These methods return a SingleIndicatorResults object, NOT a paginated iterator. Access .values to get the list of data points:

@st.cache_data(ttl=TTL_CHART, show_spinner=False)
def get_rsi(api_key: str, ticker: str, window: int, timespan: str) -> list:
    client = _get_client(api_key)
    result = client.get_rsi(ticker, params={"window": window, "timespan": timespan, "sort": "asc"})
    return result.values if result.values else []

Each item has .timestamp (ms epoch) and .value. Do NOT wrap these calls in list(islice(...)).

Multi-asset watchlist

Use list_universal_snapshots() with mixed-asset tickers in a single call:

tickers = ("AAPL", "MSFT", "X:BTCUSD", "C:EURUSD", "I:SPX")
snapshots = get_snapshot(api_key, tickers)

WebSocket trade tape (real-time panel)

Run WebSocket in a daemon thread; append messages to a session-scoped buffer:

if "trade_buffer" not in st.session_state:
    st.session_state.trade_buffer = []

# Start WebSocket thread (see bloomberg-terminal/terminal/panels/trade_tape.py)

Focus areas

multi-asset (default)

Panels: watchlist (equities + crypto + forex + indices), price chart with indicators, market status. SDK: list_universal_snapshots, list_aggs, get_sma/get_ema/get_rsi/get_macd, get_market_status(). For market status, use client.get_market_status() from the SDK. Do NOT make raw REST calls to any domain other than api.massive.com.

options

Panels: options chain table, Greeks heatmap, P&L diagram, underlying price chart. SDK: list_snapshot_options_chain, list_aggs, get_last_trade.

crypto

Panels: crypto watchlist, BTC/ETH price charts, volume comparison. SDK: list_universal_snapshots with X: tickers, list_aggs.

macro

Panels: treasury yield curve, inflation chart, labor market indicators, fed funds rate. SDK: list_treasury_yields(), list_inflation(), list_labor_market_indicators(), get_market_status(). Do NOT use the requests library or any raw HTTP calls. Use only the Massive Python SDK for all API calls. Do NOT add requests to pyproject.toml. If an SDK method is not available for a specific endpoint, note it in a comment but do not fall back to raw HTTP. Note: Economy/Federal Reserve data may require a specific plan tier. Check access at massive.com/dashboard.

Dependencies

[project]
name = "$0"
version = "0.1.0"
requires-python = ">=3.9"
dependencies = [
    "massive>=2.4.0",
    "streamlit>=1.41.0",
    "plotly>=5.24.0",
    "pandas>=2.2.0",
    "numpy>=2.0.0",
    "python-dotenv>=1.0.0",
]

Streamlit config

.streamlit/config.toml:

[theme]
base = "dark"

[server]
headless = true

Steps

1. Create the project directory structure
2. Write config.py with color palette and TTL constants
3. Write data.py with cached API call wrappers for the chosen focus area
4. Write charts.py with Plotly rendering functions
5. Write panel modules for the chosen focus area
6. Write streamlit_app.py with layout, sidebar, and panel composition
7. Write pyproject.toml, .env.example, .gitignore, .streamlit/config.toml, README
8. Provide quickstart:

   cd $0
   cp .env.example .env
   # Add your Massive API key to .env
   uv sync
   uv run streamlit run streamlit_app.py
   

9. Note: Dashboards use Streamlit (Python only). The user will need at least a Starter plan for snapshots and WebSocket data.