yummy-research-mcp

An MCP server that exposes overseas and domestic macro/market statistics as tools. The first data set consists of market sentiment/valuation indices, but it is designed to add KRX, DART, Bank of Korea ECOS, FRED, etc., in the same pattern in the future.

Current Tools

Each tool returns a JSON in the format { name, source, latest, series, ... }.

Related MCP server: rays-mcp

Development Environment

Based on uv.

cd ~/workspace/yummy-research-mcp

uv sync                          # 의존성 설치 (.venv 자동 생성)
uv run yummy-research-mcp        # MCP stdio 서버 실행
uv run python -m yummy_research_mcp.sources.cnn        # 단독 페치 디버깅
uv run pytest                    # 라이브 엔드포인트 스모크 테스트

Claude Code / Claude Desktop Registration

{
  "mcpServers": {
    "yummy-research": {
      "command": "uv",
      "args": [
        "--directory",
        "/Users/yeom/workspace/yummy-research-mcp",
        "run",
        "yummy-research-mcp"
      ]
    }
  }
}

Adding New Data Sources

1. Write a fetcher in src/yummy_research_mcp/sources/<source>.py — pure function, returns a JSON-serializable dict.

2. Add Tool + callable to the TOOLS registry in src/yummy_research_mcp/server.py.

3. Add live smoke tests in tests/.

Directory

src/yummy_research_mcp/
  __init__.py
  http.py                # 공통 urllib 래퍼 (browser-like UA / Accept-Language)
  server.py              # MCP stdio 서버 + 도구 레지스트리
  sources/
    cnn.py               # CNN Fear & Greed (production.dataviz.cnn.io)
    indexergo.py         # indexergo.com (인라인 ECharts JSON 파싱)
tests/
  test_fetchers.py

Data Source Notes

• CNN F&G: production.dataviz.cnn.io/index/fearandgreed/graphdata JSON API. Browser UA + Origin: edition.cnn.com + Referer are required (returns 418 if missing).

• indexergo: Extract the first series.data from the page's inline ECharts option JSON using balanced bracket scanning. Static HTML parsing was adopted as it is more stable than the site's /ajaxMakeChart POST endpoint.