Skip to content

Repository files navigation

Financial Intelligence Agent

Powered by Tavily & Arcade

A web-based financial research assistant. Research companies, analyze markets, create Google Docs briefings, and share to Slack — all through one chat interface.

Financial Intelligence Agent

Built with Tavily for real-time research, Arcade as the MCP runtime, and Vercel AI SDK for streaming chat with MCP tool support.


How It Works

┌──────────────────────────────────────────────────────────────┐
│                   Arcade MCP Runtime                         │
│                                                              │
│  Tavily (6 tools)   Google Docs (13 tools)   Slack (10 tools)│
│  search, research,  create, edit, share      send, read,     │
│  extract, crawl,    search, comment          list, invite    │
│  map, skill                                                  │
└──────────────────────────────────────────────────────────────┘
                           ▲
                           │  streamable HTTP + headers auth
                           │
                   ┌───────┴────────┐
                   │  Next.js App   │
                   │  Vercel AI SDK │
                   │  + Anthropic   │
                   └────────────────┘

The Vercel AI SDK's createMCPClient connects to the Arcade MCP Gateway over streamable HTTP. All 29 tools from 3 MCP servers (Tavily, Google Docs, Slack) are available through a single gateway URL. Authorization, execution, and governance are handled by Arcade.


Prerequisites

  • Node.js 18+
  • A Tavily account (for the remote MCP URL)
  • An Arcade account (free tier)
  • An Anthropic API key

Setup

Step 1: Get your Tavily MCP URL

Go to tavily.com > Overview > Generate MCP Link.

Get Tavily MCP URL

Step 2: Add Tavily as a Remote MCP Server in Arcade

Go to Arcade Dashboard > Servers > Add Server > Remote MCP. Paste the Tavily MCP URL.

Add Tavily server

Step 3: Verify Tavily tools

Arcade discovers 6 Tavily tools: tavily_search, tavily_extract, tavily_crawl, tavily_map, tavily_research, tavily_skill.

Tavily tools

Step 4: Create an MCP Gateway

Go to MCP Gateways > Create Gateway. Select Tavily + GoogleDocs + Slack. Set the auth mode to Arcade Headers.

Gateway

Copy the gateway URL.

Step 5: Configure and run

cp .env.example .env.local
# Edit .env.local with your keys
npm install
npm run dev

Open http://localhost:3000.


Environment Variables

Variable Description
ANTHROPIC_API_KEY Anthropic API key
ARCADE_GATEWAY_URL Arcade MCP Gateway URL
ARCADE_API_KEY Arcade API key (from dashboard sidebar)
ARCADE_USER_ID Your email for Arcade user context

Project Structure

app/
  page.tsx                        Chat UI with useChat, streaming, notes panel
  api/chat/route.ts               Backend: MCP client + streamText
  layout.tsx                      Root layout and metadata
  globals.css                     Tailwind + finance theme styles
  components/
    tool-progress.tsx             Tool call display with service labels
    markdown-content.tsx          Rendered markdown with react-markdown
    contextual-suggestions.tsx    Smart suggestion pills above input
    notes-panel.tsx               Collapsible right sidebar for pinned notes
lib/
  prompts.ts                      Financial analyst system prompt
images/                           Setup screenshots for this guide
.env.example                      Environment template

Example Prompts

  • "What happened with NVDA in the last 24 hours?"
  • "Research TSLA: earnings, analyst notes, SEC filings"
  • "Compare MSFT and GOOGL on their AI investments"
  • "Prepare a pre-market brief for AAPL, AMZN, META. Create a Google Doc."
  • "Post a summary of today's semiconductor news to #investment-team on Slack"

Built With

  • Tavily — Real-time search, research, and extraction
  • Arcade — The MCP runtime: authorization, reliable tools, governance
  • Vercel AI SDK — Streaming AI with MCP tool support

About

Pre-Market Intelligence Agent powered by Arcade & Tavily

Resources

Stars

3 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages