feat: Add MCP tool annotations for improved LLM integration

[bryankthompson]

Summary

Add ToolAnnotations to all 11 tools to help LLM clients understand tool behavior and make better decisions about tool selection and user confirmations.

Changes

• SDK Upgrade: mcp[cli]>=1.3.0 → mcp[cli]>=1.8.0 (required for annotations)
• Added annotations to all 11 tools in server.py

Tool Annotations Added

Tool	title	readOnlyHint	destructiveHint	openWorldHint
get_income_statements	Get Income Statements	✅	❌	✅
get_balance_sheets	Get Balance Sheets	✅	❌	✅
get_cash_flow_statements	Get Cash Flow Statements	✅	❌	✅
get_current_stock_price	Get Current Stock Price	✅	❌	✅
get_historical_stock_prices	Get Historical Stock Prices	✅	❌	✅
get_company_news	Get Company News	✅	❌	✅
get_available_crypto_tickers	Get Available Crypto Tickers	✅	❌	✅
get_crypto_prices	Get Crypto Prices	✅	❌	✅
get_historical_crypto_prices	Get Historical Crypto Prices	✅	❌	✅
get_current_crypto_price	Get Current Crypto Price	✅	❌	✅
get_sec_filings	Get SEC Filings	✅	❌	✅

Annotation Rationale

• readOnlyHint=true: All tools are read-only data retrieval operations
• destructiveHint=false: No tools modify any data
• openWorldHint=true: All tools call the external Financial Datasets API

Why This Matters

Tool annotations are part of the MCP specification that help AI clients:

• Display appropriate confirmation dialogs for destructive operations
• Make better decisions about autonomous tool execution
• Show users accurate information about what tools do

Testing

• ✅ Python import test passes
• ✅ All 11 tools verified

🤖 Generated with Claude Code