Overview
gotoHuman MCP Server enables AI agents and agentic workflows to request human approvals in a fully managed, asynchronous loop. It includes a customizable approval UI and built-in authentication, webhooks, notifications, and team features, plus an evolving training dataset to improve the approval process over time. You can trigger human approvals from your AI workflows via MCP, or install the server in your IDE to simplify integration. Installation is via npx @gotohuman/mcp-server; obtain an API key from app.gotohuman.com to configure approvals. For development, you can install dependencies, build the server, and run an MCP inspector for testing. The README also describes how to run the server locally in MCP clients such as Cursor, Claude, or Windsurf, using a sample mcpServers configuration. The Tools section provides endpoints to list forms, fetch a form schema, and request a human review using a form—returning a link to the review in the gotoHuman inbox. Together, these features make it easy to embed human-in-the-loop approvals into AI workflows and IDE-based integrations.
Features
Customizable approval UI
A configurable user interface for human approvals to suit various workflows.
Built-in authentication
Authentication mechanisms built into the MCP server for secure access.
Webhooks
Webhook support to notify external systems when approvals are requested or completed.
Notifications
Notify relevant parties about approvals and status updates.
Team features
Team-related capabilities to manage users and approvals collaboratively.
List forms
Endpoint to list all available review forms with high-level field info.
Get form schema
Endpoint to fetch the schema for a specific form to guide review requests.
Request human review with form
Endpoint to submit a human review using a specified form and return the review link.
Who Is This For?
- AI developers:Integrate MCP-based human approvals into AI workflows.
- Engineering teams:Coordinate approvals within agentic workflows and IDE integrations.
