MCP server that lets AI workflows request human approvals through gotoHuman’s fully-managed human-in-the-loop platform.
https://github.com/gotohuman/gotohuman-mcp-serverBuilding AI agents that make real-world decisions? You know the drill - at some point, you need human approval for critical actions. But rolling your own approval system means building forms, managing state, handling notifications, and dealing with user management. That's weeks of work before you even get to the interesting parts.
The gotoHuman MCP server solves this by giving your AI workflows instant access to a fully-managed human approval system. Request reviews from your agents, get approvals through a polished UI, and continue execution based on human decisions - all without building approval infrastructure.
Here's what you get out of the box:
Immediate Integration: Add to Cursor, Claude, or Windsurf with a single MCP server config. Your AI can request human approvals in the same conversation where it's planning deployments or analyzing data.
Production-Ready UI: Custom approval forms with field validation, file attachments, and team assignment. No more sending "approve Y/N" Slack messages for critical decisions.
Async by Design: Your workflows don't block waiting for approvals. Webhook callbacks let you continue processing when humans respond, whether that's minutes or hours later.
Team Collaboration: Assign reviews to specific team members, track approval history, and maintain audit trails without building user management systems.
Database Migrations: Your AI analyzes schema changes and generates migration scripts, but requires DBA approval before execution. The agent requests review with the full migration plan, diff, and risk assessment pre-filled in the approval form.
// AI requests approval with context
await requestHumanReview({
formId: 'database-migration',
fieldData: {
migration_sql: generatedMigration,
affected_tables: affectedTables,
risk_assessment: riskAnalysis,
rollback_plan: rollbackSteps
}
});
Content Publishing: AI-generated blog posts, social media content, or marketing copy gets human review before going live. Include the content, target audience analysis, and brand compliance checks.
Financial Decisions: Expense approvals, budget allocations, or procurement decisions that exceed thresholds. The AI provides cost analysis and recommendations, humans make the final call.
Code Deployments: AI suggests deployment strategies for complex releases, but requires human sign-off for production changes that affect critical systems.
The MCP integration means your AI tools already know how to use it. Whether you're building agents in Python, TypeScript, or using cursor for assisted development, the approval requests happen naturally in the same context where decisions are being made.
Setup takes under 5 minutes:
{
"mcpServers": {
"gotoHuman": {
"command": "npx",
"args": ["-y", "@gotohuman/mcp-server"],
"env": {
"GOTOHUMAN_API_KEY": "your-api-key"
}
}
}
}
Your AI immediately gains three new capabilities: listing available approval forms, fetching form schemas, and requesting human reviews with structured data.
Instead of spending cycles on approval workflows, user interfaces, and notification systems, you're building the AI logic that actually moves your business forward. The human oversight becomes a simple function call in your agent's decision tree.
Get your API key at app.gotohuman.com and add human judgment to your AI workflows today.