mcp-tool-servers

Three FastMCP HTTP servers exposing shared tools as MCP endpoints. Each runs on its own port with streamable HTTP transport.

Servers

web-search (port 8010)

Tool	Description
tavily_quick_search(query, max_results)	Quick web search via Tavily API. Returns synthesized answers and snippets.
firecrawl_deep_scrape(url)	Deep scrape a specific URL for full markdown content via Firecrawl.

finance-data (port 8011)

Tool	Description
get_ticker_data(ticker)	Market data, P/E ratios, company summary via yfinance. Supports .NS/.BO for Indian stocks.
get_bse_nse_reports(ticker)	Raw quarterly/yearly income statements, balance sheets, cash flow in markdown.

vector-db (port 8012)

Tool	Description
check_in_vector_db(identifier, index_name)	Check if documents exist in a Pinecone index.
upsert_to_vector_db(data, metadata_json, index_name)	Upsert a document to the vector DB.
retrieve_from_vector_db(query, index_name, filter_key, filter_value, top_k)	Semantic search over a Pinecone index.
add_financial_reports_to_db(ticker)	Fetch financial reports and store in the financial-reports index.
check_papers_in_db(query)	Check if relevant papers exist in the research-papers index.
retrieve_papers(query, top_k)	Semantic search over stored research papers.
download_and_store_arxiv_papers(query, max_results)	Search arXiv, download PDFs, convert to markdown, upsert to vector DB.

Structure

mcp-tool-servers/
├── pyproject.toml
├── servers/
│   ├── web_search.py           # port 8010
│   ├── finance_data.py         # port 8011
│   └── vector_db_server.py     # port 8012
├── shared/
│   ├── config.py               # Port assignments
│   └── vector_db.py            # Consolidated VectorDB class
└── scripts/
    └── run_all.py              # Launch all servers for dev

Running

All servers:

infisical run -- python scripts/run_all.py

Individual server:

infisical run -- python servers/web_search.py

Environment Variables

Variable	Used by	Description
TAVILY_API_KEY	web-search	Tavily API key
FIRECRAWL_API_KEY	web-search	Firecrawl API key
PINECONE_API_KEY	vector-db	Pinecone API key
NVIDIA_API_KEY	vector-db	NVIDIA embeddings API key
GEMINI_API_KEY	vector-db (if using gemini provider)	Google AI API key

Dependencies

fastmcp>=3.0, tavily-python, firecrawl-py, yfinance, arxiv, pymupdf4llm, pinecone, langchain-google-genai, langchain-nvidia-ai-endpoints, langchain-text-splitters, python-dotenv