setup - SKILL.md Agent Skill

name: setup description: Walk through initial setup and authentication for this Daloopa starter kit

Walk the user through setting up this Daloopa starter kit step by step. Be conversational and helpful.

Step 1: Verify Claude Code

Confirm Claude Code is running (if the user is seeing this, it is — tell them they're good).

Step 2: Install Python Dependencies

Check if required packages are installed. Offer to install them:

pip3 install -r requirements.txt

This installs: requests, beautifulsoup4, html2text, yfinance, openpyxl, python-docx, docxtpl, matplotlib, fredapi.

These are needed for market data, chart generation, Excel model building, and Word document rendering.

Step 3: Daloopa Authentication

Ask the user which authentication method they'd like to use:

Option A: OAuth (Recommended)

The .mcp.json is already configured for OAuth
On the next MCP tool call, a browser window will open for Daloopa login
No additional configuration needed
Just make sure they have a Daloopa account at daloopa.com

Option B: API Key

Ask the user for their Daloopa API key
Create/update .env with their key: DALOOPA_API_KEY=<their_key>
Update the daloopa entry in .mcp.json to include the API key header (keep the daloopa-docs entry as-is):

{
  "mcpServers": {
    "daloopa": {
      "type": "http",
      "url": "https://mcp.daloopa.com/server/mcp",
      "headers": {
        "x-api-key": "${DALOOPA_API_KEY}"
      }
    },
    "daloopa-docs": {
      "type": "http",
      "url": "https://docs.daloopa.com/mcp"
    }
  }
}

Tell them they'll need to restart Claude Code for the change to take effect

Step 4: Optional API Keys

Ask if they want to configure optional API keys for enhanced functionality:

FRED API Key (recommended for DCF/valuation work):

Free at https://fred.stlouisfed.org/docs/api/api_key.html
Used for risk-free rate in WACC calculations
Without it, a default rate of 4.5% is used
Add to .env: FRED_API_KEY=<their_key>

Step 5: Verify MCP Connection

This project connects to two Daloopa MCP servers:

daloopa (mcp.daloopa.com/server/mcp) — Financial data (fundamentals, KPIs, SEC filings)
daloopa-docs (docs.daloopa.com/mcp) — Daloopa knowledgebase (API docs, how-tos, usage help)

Run a quick test by calling discover_companies with a well-known ticker like "AAPL" to confirm the data MCP server is connected and responding. Show the user the result.

Step 6: Verify Market Data

If the user has a market data MCP configured (e.g., a financial data provider with stock quote tools), test it by looking up AAPL.

If no market data MCP is available, fall back to the infra script: python infra/market_data.py quote AAPL This should return current price, market cap, etc.

Step 7: Create Word Template

Run: python scripts/create_template.py This creates the research note template at templates/research_note.docx.

Step 8: Quick Tour

Tell the user about the available slash commands:

Building Block Skills (markdown reports):

/earnings TICKER — Full earnings analysis with guidance tracking
/tearsheet TICKER — Quick one-page company overview
/industry TICKER1 TICKER2 ... — Cross-company comparison
/bull-bear TICKER — Bull/bear/base scenario framework
/guidance-tracker TICKER — Track management guidance accuracy
/inflection TICKER — Auto-detect metric accelerations/decelerations
/capital-allocation TICKER — Buybacks, dividends, shareholder yield
/dcf TICKER — DCF valuation with sensitivity analysis
/comps TICKER — Trading comparables with peer multiples

Investment Deliverables (.docx, .xlsx, .pdf):

/research-note TICKER — Professional Word research note
/build-model TICKER — Multi-tab Excel financial model
/initiate TICKER — Both research note + Excel model (initiating coverage)
/update TICKER — Refresh existing coverage with latest data
/ib-deck TICKER — Institutional-grade pitch deck (HTML → PDF)

All output is saved to the reports/ directory.

Suggest they try /tearsheet AAPL as a quick first test to see everything working end-to-end.