Fintools public MCP connector and paid Fintools MCP routing shell for market-analysis toolbelt
Fintools MCP has 17 tools with inconsistent quality. Public tools (4) have reasonable descriptions and basic schemas. Paid tools (13) are stubs returning upgrade_required errors with no real implementation. Naming is verb-forward (get_, check_, screen_, find_, analyze_) but descriptions vary widely, some are 10-15 chars (too short), others are adequate. No parameter descriptions for most paid tools. No output schemas documented. Error handling is minimal: paid tools return a generic upgrade_required JSON, not actionable recovery guidance. The server conflates two product tiers in one codebase, making it difficult to assess true capability.
Explain the public connector and the paid Fintools MCP package. Use this when a user asks what Fintools can do, what package is installed, how to upgrade, or why a market-analysis workflow is unavailable here.
Paid Fintools MCP required: options-chain context and liquidity filters.
Paid Fintools MCP required: trade-stat calculations and review context.
Paid Fintools MCP required: ATR-based position sizing.
Paid Fintools MCP required: risk-based position sizing.
Confirm that the public Fintools connector is connected and responding.
13 of 17 tools are stubs with no real implementation. Paid tools return upgrade_required errors instead of actual functionality, making them non-functional placeholders. This violates the single-responsibility principle and confuses agents about what is actually available.
Paid tool descriptions are extremely short (20-35 chars) and lack actionable context. 'Paid Fintools MCP required: <capability>' tells the agent nothing about what the tool does, when to use it, or what parameters control. Descriptions must be 50-200 chars with clear intent.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
Paid Fintools MCP required: side-by-side technical comparison.
Paid Fintools MCP required: breakout discovery and candidate surfacing.
Show public connector cache status and provider request counters.
Paid Fintools MCP required: day-context gates, flags, and fresh_add_status.
Paid Fintools MCP required: option quote, spread, IV, and entry P&L context.
Get a basic stock quote in the public Fintools connector. Args: ticker: Stock symbol, such as AAPL, SPY, or TSLA.
Paid Fintools MCP required: support/resistance levels and structure.
Paid Fintools MCP required: RSI, MACD, ATR, EMAs, Fibonacci, and chart context.
Paid Fintools MCP required: graduated trend score with component breakdowns.
Paid Fintools MCP required: custom screens by trend, RSI, EMAs, and volume.
Paid Fintools MCP required: winner-similarity candidate ranking and triage.
No output schemas documented for any tool. Agents cannot plan downstream calls or extract structured data. Even public tools (get_stock_quote, get_data_source_stats) lack documented return types. This violates the response-shaper pattern.
Paid tool parameters lack descriptions. screen_stocks has 10 params (rsi_max, rsi_min, trend_min, etc.) with no guidance on valid ranges, units, or when to use each. LLMs cannot reason about parameter selection without descriptions.
Error handling is generic and non-actionable. All paid tools return the same upgrade_required JSON with no recovery guidance. Agents cannot distinguish between retryable errors, user-fixable issues, or fatal failures. Error responses should guide the agent's next step.