Tiger Brokers MCP (Model Context Protocol) System - Connect Claude AI to Tiger Brokers API for trading and market data operations
Tiger MCP has moderate definition quality with clear naming and reasonable parameter documentation, but lacks structured output schemas and contains several parameter completeness gaps. All 6 tools follow verb_noun naming convention (tiger_list_accounts, tiger_add_account, tiger_get_quote, tiger_get_kline, tiger_get_contracts, tiger_get_financials). Descriptions are present and reasonably detailed (100-200 chars typical). However, tool output schemas are not documented in the source code provided, only input schemas are visible. Parameter descriptions are generally present but some lack detail on constraints, formats, and valid ranges. Error handling patterns are not evident in the provided source.
Add a new Tiger account to the system. Creates a new Tiger account configuration with encrypted credential storage, proper validation, and initial token setup for API access.
Get detailed contract information for symbols. Retrieves comprehensive contract details including security type, exchange, currency, lot size, and other trading specifications needed for order placement and analysis.
Get financial data and key metrics for a symbol. Retrieves fundamental financial information including balance sheet, income statement, cash flow data, and key financial ratios useful for investment analysis and valuation.
Get historical K-line (candlestick) data for a symbol. Retrieves historical price data including OHLCV (Open, High, Low, Close, Volume) for technical analysis and charting purposes.
Get real-time quote for a symbol. Retrieves current market data including bid/ask prices, last trade, volume, and other real-time market information for a specified symbol.
Output schemas not documented. Tools return data but no schema documentation visible for what fields, types, and structure LLMs should expect from responses. This forces LLMs to guess output shape, increasing hallucination risk and breaking downstream tool chaining.
tiger_add_account exposes secrets as parameters. The 'api_key' and 'secret_key' parameters accept raw credentials as input. These will be logged in agent traces, prompt history, and audit logs, creating a security vulnerability. Credentials must be injected server-side via environment variables or vault.
Parameter constraint details missing. Parameters like 'period' in tiger_get_kline and 'account_type' in tiger_list_accounts have enums documented in descriptions, but lack explicit enum constraints in JSON Schema. LLMs may hallucinate invalid values. Parameters like 'count' (max 300) lack min/max constraints.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 50 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 21 | - | v1 |
List all configured Tiger accounts with their status. Retrieves a comprehensive list of all Tiger accounts configured in the system, including their current status, permissions, token validity, and other metadata.
Pagination not visible. tiger_list_accounts returns 'a comprehensive list' but no limit, offset, or cursor parameters are documented. Large account lists could exhaust context windows. No pagination guidance provided.
Error handling guidance absent. No documentation of what errors these tools return, how they are classified (retryable, user-fixable, fatal), or what recovery steps the LLM should take. A tool calling external data services should document timeout behavior and partial failure modes.