Model Context Protocol (MCP) Server

The RiskInMind MCP Server exposes the complete suite of RiskInMind AI risk and lending endpoints as standard MCP tools. This enables AI assistants (such as Claude Desktop, Cursor, Pi, and Zed) to autonomously execute credit risk evaluations, generate Fannie Mae Form 1003 PDFs, run CRE loan models, inspect uploaded documents for tampering, and synthesize financial advisory memos.

Connection Details

RiskInMind provides a hosted, remote HTTP MCP endpoint as well as local stdio execution:

Type	Endpoint / Command	Authentication
Remote HTTP (SSE / Streamable)	https://mcp.riskinmind.ai/mcp	x-api-key: rim_<api_key> or RISKINMIND_API_KEY
Local Stdio (Node.js)	node src/stdio.js (inside riskinmind-mcp)	RISKINMIND_API_KEY env var

---

Client Setup

1. Claude Desktop

Add the RiskInMind server configuration to your claude_desktop_config.json:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "riskinmind": {
      "command": "npx",
      "args": ["-y", "riskinmind-mcp"],
      "env": {
        "RISKINMIND_API_KEY": "rim_your_api_key_here",
        "RISKINMIND_BASE_URL": "https://api.riskinmind.ai/api/v1"
      }
    }
  }
}

Or connect directly to the remote HTTP MCP server:

{
  "mcpServers": {
    "riskinmind-remote": {
      "url": "https://mcp.riskinmind.ai/mcp",
      "headers": {
        "x-api-key": "rim_your_api_key_here"
      }
    }
  }
}

---

2. Cursor IDE

In your project root, create or update .cursor/mcp.json:

{
  "mcpServers": {
    "riskinmind": {
      "url": "https://mcp.riskinmind.ai/mcp",
      "headers": {
        "x-api-key": "rim_your_api_key_here"
      }
    }
  }
}

---

3. Pi Coding Agent Harness

Add to ~/.pi/agent/mcp.json or your project settings:

{
  "mcpServers": {
    "riskinmind": {
      "url": "https://mcp.riskinmind.ai/mcp",
      "headers": {
        "x-api-key": "rim_your_api_key_here"
      }
    }
  }
}

---

Available MCP Tools

The RiskInMind MCP server provides 20 specialized banking and risk tools:

Consumer Loans & Underwriting

Tool Name	Parameters	Description
predict_loan	age, income, loanAmount, creditScore, monthlyExpenses, numberOfCreditLines, interestRate, employmentLength, debtToIncomeRatio, loanPurpose, propertyValue	Predict loan approval probability, default likelihood, and risk tier
calculate_cecl	portfolioBalance, historicalLossRate, macroeconomicAdjustment, weightedAverageMaturity, reversionPeriodMonths	Calculate Current Expected Credit Loss (CECL) allowance and reserves
generate_1003_application	borrowerDetails, loanDetails, employmentDetails, assetsLiabilities	Generate standardized Fannie Mae / Freddie Mac Form 1003 PDF
generate_loan_report	loanId or applicantDetails	Generate comprehensive credit evaluation and decision memo PDF
create_loan	loanPayload	Create a new loan record in the institution database
get_loan	id	Retrieve detailed loan application record by ID

---

Commercial Real Estate (CRE)

Tool Name	Parameters	Description
predict_cre_loan	propertyType, loanAmount, netOperatingIncome, capRate, dscr, ltv, debtYield, occupancyRate	Run probabilistic CRE loan risk and underwriting model
predict_cre_loan_pdf	propertyType, propertyDetails, financials	Generate executive CRE risk memorandum PDF

---

Credit Card Analytics

Tool Name	Parameters	Description
predict_credit_card_loan	annualIncome, creditScore, existingCreditLimit, utilizationRate, delinquencyHistory	Decision credit card applications and recommended credit limits

---

Document Forensics & Tampering

Tool Name	Parameters	Description
check_document_tampering	fileBase64, fileName, mimeType	Detect digital image/PDF manipulation, altered font layers, and metadata forgery
verify_document	documentId	Verify document authenticity status against institutional records

---

Financial Document Processing & Reports

Tool Name	Parameters	Description
generate_bank_statement_report	fileBase64, fileName	Extract cash flow, recurring deposits, and NSF alerts from bank statement PDFs
generate_tax_report	fileBase64, fileName	Analyze Form 1040 / W-2 / 1099 tax returns and compute adjusted gross income
generate_credit_report	fileBase64, fileName	Parse credit bureau PDF statements and summarize trade lines

---

Document Generation & Voice Memos

Tool Name	Parameters	Description
generate_document	type ("nda" | "loan_agreement" | "offer_letter"), variables	Generate signed legal PDFs from institutional templates
polish_voice_memo	rawTranscript	Format meeting or phone call transcript into clean Markdown
generate_voice_memo_report	rawTranscript, officerName, borrowerName	Generate official loan committee discussion memo PDF

---

Virtual AI Banking Agents

Tool Name	Parameters	Description
chat_bot_ava	message, context	Public frontline virtual assistant for customer support
chat_bot_erina	message, context	Compliance & regulatory policy guidance agent
chat_bot_david	message, context	Credit underwriting & risk assessment copilot
chat_bot_sean	message, context	Commercial loan structuring advisor
chat_bot_mark	message, context	Portfolio executive intelligence & CECL strategist

---

Example AI Prompts

Once connected to your MCP client, you can run prompts such as:

Evaluate a $450,000 residential loan application for a borrower with:
- Annual Income: $145,000
- Credit Score: 720
- Monthly Expenses: $3,200
- Debt-to-Income: 32%
- Loan Purpose: Home Purchase

Use `predict_loan` to compute the risk tier, then format the recommendation.

Inspect this uploaded bank statement PDF for any digital tampering or font layer alterations using `check_document_tampering`.