Finch MCP

Finch MCP puts Finch's unified employment API behind the Model Context Protocol. Finch normalizes data from a large network of US HRIS and payroll providers into a single schema, and the MCP server hands that normalized schema to an agent so it can answer employment questions without a separate integration per provider. The exposed surface follows Finch's data model. An agent can read company level information about a connected employer, pull the directory of employees and contractors, and open an individual record for profile and employment detail such as job title, department, manager, employment status and compensation structure. Payroll is the deeper part: pay statement level data with earnings, taxes, employee deductions and employer contributions standardized across providers, alongside payroll run metadata. Where the underlying provider supports it, write back lets an agent create and manage deductions and contributions rather than only reading them. The audience is teams building benefits, retirement, insurance, HR automation or fintech products on top of employer payroll connections, plus internal operations teams reconciling headcount against payroll. The data involved is sensitive personal and compensation information, so access control and where inference actually happens deserve more scrutiny here than with most connectors. Finch distributes the server as an installable package.