Stock Intelligence MCP

What are MCP Apps? An extension to the Model Context Protocol that lets MCP servers return interactive HTML interfaces — charts, forms, dashboards — rendered directly inside the AI conversation. Announced in the MCP Apps blog post, supported in Claude Desktop, ChatGPT, VS Code, and more. This project uses MCP Apps to render canvas-based financial charts with inline controls that let you adjust parameters without a new AI call.

How It Works

Two modes — visual charts or inline data:

• "Show me" → AI uses stock_app → interactive chart with controls appears

• "What's the price of" → AI uses stock_quote → numbers in text

You: "Show me NVIDIA's stock chart for the last year"

AI calls: stock_app(symbol="NVDA", view="chart", period="1y")

→ Interactive candlestick chart with volume bars and hover crosshair.
  Inline controls: switch between 1M/3M/6M/1Y/2Y/5Y and Candle/Line/Area
  without making a new AI call.

You: "Show me Apple's quarterly financials"

AI calls: stock_app(symbol="AAPL", view="financials")

→ Column chart of revenue, net income, gross profit.
  Inline controls: switch between Income/Balance Sheet/Cash Flow,
  Quarterly/Annual, and Column/Bar/Dual Axes/Table.

You: "What's Tesla's price right now?"

AI calls: stock_quote(symbol="TSLA")

→ Text: "TSLA is trading at $248.50, up +2.3% today.
   Market cap $789B, 52-week range $152–$299."

You: "Is Microsoft undervalued?"

AI calls: stock_app(symbol="MSFT", view="valuation")

→ DCF gauge chart: intrinsic value vs current price, margin of safety.
  Verdict: undervalued/overvalued/fairly valued.

You: "Who's reporting earnings this week?"

AI calls: stock_earnings_calendar(days_ahead=7)

→ Text: "47 earnings reports this week: AAPL (Mon), GOOGL (Tue)..."

Related MCP server: FinImpulse MCP Server

Quick Start

1. Double-click the .mcpb file

Open stock-intelligence-mcp.mcpb — Claude Desktop will prompt you to install it.

2. Add your FMP API key

Get a free key (30 seconds): https://site.financialmodelingprep.com/developer/docs

Free tier: 250 API calls/day — enough for a full day of research. Claude Desktop will ask for it during setup.

3. Start asking

"Show me AAPL's stock chart as a line chart over 2 years"

"What are the latest analyst ratings for NVDA?"

"Show me Apple's balance sheet — annual view"

"Is Tesla undervalued or overvalued right now?"

"What's Amazon's earnings beat/miss record?"

"Who's reporting earnings this week?"

"What's the price of Google right now?"

"Show me today's market movers as a heatmap"

"Compare Apple's quarterly cash flow"

10 Tools

stock_app — Interactive Charting (1 tool, 7 views)

The AI picks the view based on your question. Each chart has inline controls to adjust parameters without a new AI call.

Inline Data Tools (7 tools — text responses)

Use these when you want numbers in text, not a visual chart. Every data tool has a chart equivalent.

Utility Tools (2 tools)

Architecture

stock_app (charting)                    Data tools (inline text)
────────────────────                    ────────────────────────
User asks question                      User asks for numbers
  → AI picks view + params               → AI picks data tool
  → Tool fetches from FMP API            → Tool fetches from FMP API
  → Returns data with chart_type         → Returns structured data
  → MCP App renders interactive chart    → AI formats as text response
  → User adjusts with inline controls
  → Controls re-call stock_app

All charts are rendered with inline canvas — no external JS libraries. The HTML file is self-contained (~55KB) with 19 chart renderers.

Data Source

All data comes from Financial Modeling Prep (FMP) — free tier only:

Free tier: 250 API calls/day. One API key, takes 30 seconds to get one.

Rebuild the .mcpb

./build-mcpb.sh

Outputs a fresh stock-intelligence-mcp.mcpb. Double-click to reinstall.

Alternative: Run from source

pip install -e .

export FMP_API_KEY=your_key_here

stock-intelligence-mcp

Then add to your Claude Desktop config (~/.claude/claude_desktop_config.json):

{
  "mcpServers": {
    "stock-intelligence": {
      "command": "stock-intelligence-mcp",
      "env": {
        "FMP_API_KEY": "your_key_here"
      }
    }
  }
}

Hosted Version

Don't want to manage API keys or run locally?

mcpbundles.com — same tools, zero setup, 200+ FMP tools always available.

Contributing

PRs welcome — new chart types, data visualizations, UI improvements.

git clone https://github.com/thinkchainai/stock-intelligence-mcp.git
cd stock-intelligence-mcp
pip install -e .

License

MIT — see LICENSE.

Built by MCPBundles.