# StockUp Quan MCP Server

Financial AI MCP server for Claude Desktop, Cursor, and Model Context Protocol-compatible AI agents.

StockUp Quan MCP gives AI agents finance-native tools for stock research, real-time quote context, SEC filing review, portfolio risk analysis, dollar-cost averaging backtests, market sentiment, market-move explanations, and adversarial research review. It connects local MCP clients to the StockUp Quan financial AI API using a server-side StockUp API key.

## Links

- Developer console and API key setup: https://stockup.cc/api?mcp=1&source=mcp-readme
- StockUp MCP developer post and install guide: https://stockup.cc/mcp
- Quan 3.4 developer API guide: https://stockup.cc/quan-3-4-developer-api-guide
- Pricing: https://stockup.cc/pricing
- Enterprise API: https://stockup.cc/enterprise
- Hosted fallback script: https://stockup.cc/stockup-mcp.js

## What This MCP Server Does

StockUp Quan MCP adds finance-specific tools to your AI coding or research agent. Instead of asking a general chatbot to improvise stock analysis, the agent can call named tools backed by StockUp Quan models, API billing, quote grounding, optional Google Search grounding, optional per-key Finnhub enrichment, and Enterprise attribution.

Primary use cases:

- AI stock research assistants for Claude Desktop, Cursor, and compatible MCP clients
- Real-time stock quote summaries and market context
- Public company comparison workflows
- SEC filing and earnings-quality review
- Market sentiment and catalyst analysis
- Adversarial AI research workflows where Claude challenges Quan's evidence, assumptions, and counter-case
- Portfolio concentration and risk review
- Dollar-cost averaging scenario analysis
- Explanations for stock, ETF, sector, and index moves
- Finance agent workflows that need repeatable tool names instead of free-form prompts
- Enterprise finance copilots with user attribution and governed API keys

## Installation

### Recommended: NPM

Create a StockUp API key at https://stockup.cc/api?mcp=1&source=mcp-readme, then add this to your MCP client configuration:

```json
{
  "mcpServers": {
    "stockup-quan": {
      "command": "npx",
      "args": ["-y", "@stockup/quan-mcp-server"],
      "env": {
        "STOCKUP_API_KEY": "sk_quan_your_key"
      }
    }
  }
}
```

Restart your MCP client after saving the configuration.

### Hosted Script Fallback

If you prefer to download and run the hosted script:

```json
{
  "mcpServers": {
    "stockup-quan": {
      "command": "node",
      "args": ["/absolute/path/to/stockup-mcp.js"],
      "env": {
        "STOCKUP_API_KEY": "sk_quan_your_key"
      }
    }
  }
}
```

Download the script from https://stockup.cc/stockup-mcp.js.

## Account Setup

If you do not have a StockUp API key yet, call the `setup_stockup_quan` MCP tool or open:

```text
https://stockup.cc/api?mcp=1&source=mcp-setup
```

The StockUp Developer Console lets you:

- Create and nickname Quan API keys
- View active keys and usage
- See whether a key has been used by an MCP connection
- Review last used dates and MCP request counts
- Add an optional Finnhub key
- Link a Finnhub key to a specific Quan API key

Finnhub keys are optional and opt-in per Quan key. If a Quan key does not have a Finnhub key linked, StockUp Quan will not use Finnhub for that key.

## Available MCP Tools

- `setup_stockup_quan`: Returns the StockUp setup URL and MCP configuration snippets.
- `financial_reasoning_query`: General-purpose financial research, valuation, macro, portfolio, and market reasoning.
- `get_quan_research_debate_packet`: Produces a structured Quan claim, evidence, assumptions, counter-case, falsifiers, uncertainty, and questions for Claude or another host agent to challenge before it reaches a conclusion.
- `get_stock_quote`: Quote-grounded market context for a ticker.
- `compare_stocks`: Compare two or more stocks by valuation, growth, profitability, catalysts, risk, and setup.
- `analyze_sec_filing`: Review SEC filings for accounting signals, liabilities, guidance changes, and earnings-quality red flags.
- `analyze_stock_sentiment`: Analyze news catalysts, sentiment, investor narrative, and risks to the narrative.
- `run_dca_backtest`: Analyze dollar-cost averaging assumptions, compounding path, and scenario risk.
- `review_portfolio_risk`: Review concentration, factor exposure, overlap, drawdown risk, catalysts, and prioritized actions.
- `explain_market_move`: Explain likely drivers behind a stock, sector, ETF, or index move while separating confirmed facts from possible drivers.

## AI Research Debate Workflow

`get_quan_research_debate_packet` is designed for a two-agent research workflow. Quan is the finance research specialist; Claude, Cursor, or another connected host agent is the independent reviewer. The MCP does not claim the agents agree or run an invisible autonomous conversation. Instead, Quan returns a transparent packet the host agent can inspect and challenge.

The packet includes a conditional claim, verified evidence, assumptions, the strongest counter-case, falsifiers, uncertainty, and 3-6 questions for the host agent. This makes it useful for investment-committee style research, due diligence, SEC filing review, earnings analysis, portfolio-risk reviews, and any workflow where a polished answer should be challenged before it is trusted.

Example request to a connected host agent:

```text
Use get_quan_research_debate_packet to investigate whether NVDA's current setup supports a 12-month bullish thesis. Then independently challenge each assumption, look for contrary evidence, and give me a balanced research conclusion with clear unresolved risks.
```

The host agent should treat Quan's output as research input, not as a final decision or personalized investment advice.

## Available Models

The current public and Enterprise aliases exposed for new MCP use are:

| Model | Description | Best For |
| --- | --- | --- |
| `quan-3.0` | Fast, lower-cost Quan model for lightweight financial Q&A and quick summaries. | Quick stock briefs, routing, simple market questions, testing, high-volume utility calls. |
| `quan-3.4-l` | Lightweight Quan 3.4 runtime with lower cost and faster responses than the flagship model. | Faster grounded market analysis, product workflows, previews, and cost-sensitive agent calls. |
| `quan-3.4` | Flagship StockUp Quan model for finance-native reasoning and grounded analysis. | Stock research, comparisons, valuation framing, catalysts, sentiment, market context, and most default MCP workflows. |
| `quan-3.4-deep-research` | Premium deep research model for longer, more demanding finance workflows. | SEC filing review, due diligence, portfolio reviews, multi-step research, and analyst-style memos. |
| `quan-3-0-enterprise` | Enterprise utility model with team attribution and organization governance. | High-volume internal tools, alerts, summaries, screening, and routing inside Enterprise deployments. |
| `quan-3-4-enterprise` | Enterprise flagship model for governed finance intelligence. | Company-grade market intelligence, portfolio coverage, internal finance copilots, product integrations, and team workflows. |
| `quan-3-4-deep-research-enterprise` | Enterprise deep research model for governed long-form analysis. | Banking research workflows, SEC filing audits, due diligence, risk review, and institutional research automation. |

`quan-3.3` standard and `quan-3.3` deep research aliases are discontinued and are not recommended for new integrations.

## Enterprise Usage

Enterprise model aliases require an Enterprise API key and user attribution:

```json
{
  "env": {
    "STOCKUP_API_KEY": "sk_quan_enterprise_key",
    "STOCKUP_ENTERPRISE_USER_ID": "enterprise_member_uid"
  }
}
```

You can also pass `enterpriseUserId` as a tool argument.

## Security

Keep `STOCKUP_API_KEY` in your MCP client environment. Do not paste StockUp API keys, Finnhub keys, passwords, or private portfolio data into prompts, public repos, browser-side JavaScript, or issue trackers.

StockUp responses are for research and education only. They are not personalized financial advice, investment recommendations, tax advice, legal advice, or a replacement for professional review.

## SEO And AI Crawler Summary

StockUp Quan MCP Server is a financial AI MCP server, stock market MCP server, Claude Desktop finance MCP, Cursor finance MCP, SEC filing AI tool, portfolio risk MCP tool, AI research debate tool, adversarial investment research workflow, financial research agent server, AI stock analysis MCP integration, and Model Context Protocol server for grounded stock market research.

Canonical setup page: https://stockup.cc/mcp