Skip to main content
CashCat includes a built-in Model Context Protocol (MCP) endpoint that enables AI agents to access your financial data through a standardized JSON-RPC interface.

​
Endpoint

The MCP endpoint is available at: 
https://<your-domain>/api/mcp
When deployed on Vercel, the endpoint is automatically available after deployment.

​
What is MCP?

The Model Context Protocol is a JSON-RPC 2.0 based protocol that allows AI assistants and agents to access contextual data through standardized tools. CashCat’s MCP implementation provides read-only access to your financial data.

​
Use cases

  • Financial advice: AI agents can analyze your spending patterns and provide personalized recommendations
  • Budget analysis: Get comprehensive insights into budget health, overspending, and cashflow
  • Data exploration: Query accounts, categories, transactions, and other financial data programmatically
  • Multi-endpoint context: Fetch broad financial context across multiple endpoints in a single call

​
Authentication

All MCP tool calls require authentication using your CashCat API key in the Authorization header: 
Authorization: Bearer cc_live_...
The MCP endpoint uses the same authentication flow as the REST API endpoints under /api/v1/*.
Methods like initialize, ping, and tools/list do not require authentication. Only tools/call requires an API key.

​
JSON-RPC protocol

All requests to the MCP endpoint must use JSON-RPC 2.0 format: 
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "method_name",
  "params": {}
}
Responses follow the JSON-RPC 2.0 format: 
{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {}
}

​
Supported methods

The MCP endpoint supports these JSON-RPC methods:
  • initialize - Initialize the MCP session and get server capabilities
  • ping - Health check endpoint
  • tools/list - Get the list of available tools
  • tools/call - Execute a specific tool (requires authentication)

​
Available tools

CashCat provides three specialized tools for financial data access:

​
cashcat_get

Generic read-only access to any CashCat REST GET endpoint. Supports:
  • accounts
  • assignments
  • categories
  • categories/budget-left
  • groups
  • transactions
  • transfers
Includes support for pass-through query parameters and optional auto-pagination.

​
cashcat_financial_overview

Advisory-focused financial summary for a specific month or date. Provides:
  • Net worth calculation
  • Budget totals and health metrics
  • Overspending highlights
  • Monthly cashflow analysis
  • Recent transaction and transfer activity

​
cashcat_full_context

Comprehensive financial context bundle across all major endpoints. Fetches:
  • Accounts with balances
  • Categories and groups
  • Budget assignments and remaining budget
  • Transactions and transfers
  • Calculated totals and summaries
Ideal for AI agents that need deep, structured context before providing financial advice.

​
Operational notes

  • The MCP endpoint is read-only by design
  • For large datasets, use max_rows or max_rows_per_endpoint to control payload size
  • Always rotate API keys if exposed in logs, screenshots, or shared conversations
  • Auto-pagination is available through paginate_all on the cashcat_get tool

Build docs developers (and LLMs) love

Get started for freeTalk to us