跳转至

Tutorial 4: MCP Tool Marketplace

Discover, search, and add new MCP tools for your research workflow.

What is the MCP Tool Marketplace?

The MCP Tool Marketplace is a registry that indexes all MCP servers in mcp_servers/. Each server provides specialized tools for:

  • Financial data (stocks, fundamentals, derivatives)
  • Macroeconomic indicators (GDP, CPI, interest rates)
  • Academic research (arXiv, NBER working papers)
  • News and market sentiment

Using MCPToolRegistry

Python API

from scripts.core.mcp_tool_market import MCPToolRegistry

# Build registry from mcp_servers/ directory
registry = MCPToolRegistry.from_directory("mcp_servers")

# Search for tools
results = registry.search("gdp", category="macro_data")
for tool in results:
    print(f"{tool.mcp_server}::{tool.name}")
    print(f"  Score: {tool.quality_score}")
    print(f"  Description: {tool.description[:80]}...")

# Get tools by category
macro_tools = registry.get_by_category("macro_data")

# Get tools by server
tushare_tools = registry.get_by_server("user-tushare")

# Generate marketplace report
report = registry.get_marketplace_report()
print(f"Total tools: {report['total_tools']}")
print(f"Top 5: {report['top_5_by_quality']}")

CLI Interface

# Show all tools
python scripts/core/mcp_tool_market.py --dir mcp_servers

# Search for specific tools
python scripts/core/mcp_tool_market.py --search "gdp" --report

# Filter by category
python scripts/core/mcp_tool_market.py --category financial

# Show tools from a specific server
python scripts/core/mcp_tool_market.py --server user-tushare

# Export as JSON
python scripts/core/mcp_tool_market.py --json > registry.json

Quality Scoring

Tools are scored 0.0–1.0 based on:

Factor Weight Description
Description 0.2 Rich, non-empty description (>10 chars)
Schema 0.2 Has input schema with properties
Real API 0.3 No mock/demo signals
Examples 0.1 Has usage examples in description
Total 1.0

Score Interpretation

Score Range Quality Level
0.8–1.0 High quality: real API, complete schema, examples
0.5–0.8 Medium quality: some documentation, may be partial
0.0–0.5 Low quality: mock, incomplete, or minimal

Tool Categories

Category Description Example Servers
financial A-share data, financial statements user-tushare, user-eastmoney
macro_data GDP, CPI, M2, interest rates user-financial, user-wb-data
market_data Quotes, yields, derivatives user-eodhd, user-enhanced-finance
academic Working papers, literature user-arxiv, user-nber-wp
utility File operations, LaTeX, code execution user-filesystem-mcp, user-latex-mcp

Adding New MCP Tools

Step 1: Create Server Directory

mcp_servers/
└── user_your_server/
    ├── SERVER_METADATA.json
    └── tools/
        ├── your_tool_1.json
        └── your_tool_2.json

Step 2: Create SERVER_METADATA.json

{
  "name": "user_your_server",
  "description": "Your custom data source for research",
  "version": "1.0.0",
  "author": "Your Name",
  "capabilities": ["stock_data", "financial_statements"],
  "requires_api_key": false
}

Step 3: Create Tool Definitions

// tools/get_stock_quote.json
{
  "name": "get_stock_quote",
  "description": "Get real-time stock quote for a given ticker. For example, params: {ticker: '000001.SZ'} returns current price, volume, and bid/ask spread.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "ticker": {
        "type": "string",
        "description": "Stock ticker code, e.g. '000001.SZ'"
      },
      "market": {
        "type": "string",
        "description": "Market: 'CN' for China, 'US' for US"
      }
    },
    "required": ["ticker"]
  }
}

Step 4: Register Server in Cursor

Add to .cursor/settings.json or Cursor MCP settings:

{
  "mcpServers": {
    "user_your_server": {
      "command": "python",
      "args": ["mcp_servers/user_your_server/server.py"]
    }
  }
}

Step 5: Rebuild Registry

python scripts/core/mcp_tool_market.py --dir mcp_servers --search "your_tool"

MCP Tool Usage Examples

Getting Macro Data

# Use via MCP
server: user-financial
tool: get_macro_china
params: { "indicator": "gdp" }

# Or via Python
from scripts.research_framework.data_fetcher import DataFetcher

fetcher = DataFetcher()
gdp_data = fetcher.get_macro_data("china_gdp")

Getting Stock Data

# A-share via Tushare (requires API key)
server: user-tushare
tool: get_daily_quote
params: { "ts_code": "000001.SZ", "start_date": "20240101" }

# US stocks via yfinance
server: user-yfinance
tool: get_financials
params: { "symbol": "AAPL" }

Getting Research Reports

server: user-eastmoney-reports
tool: get_research_report
params: { "ts_code": "000001.SZ", "max_results": 20 }

Marketplace Report Example

{
  "total_tools": 142,
  "total_servers": 25,
  "by_category": {
    "financial": 38,
    "macro_data": 45,
    "market_data": 28,
    "academic": 12,
    "utility": 19
  },
  "category_avg_quality": {
    "market_data": 0.72,
    "macro_data": 0.68,
    "financial": 0.65,
    "academic": 0.55,
    "utility": 0.45
  },
  "requires_api_key": 8,
  "mock_tools": 15,
  "top_5_by_quality": [
    {"server": "user-wb-data", "name": "get_wb_indicator", "score": 0.95},
    {"server": "user-eodhd", "name": "get_ust_yield_rates", "score": 0.92},
    ...
  ]
}

Next Steps