Cursor / Windsurf / IDE Agents + RentAHuman

Add RentAHuman to Cursor, Windsurf, and other AI-powered IDEs via MCP. Let your coding assistant hire humans for real-world tasks without leaving your editor.

IDE Agent Integration: Cursor, Windsurf, and MCP-Enabled Editors

Modern AI-powered IDEs like Cursor and Windsurf support the Model Context Protocol (MCP), allowing you to extend your coding assistant with external tools. By adding the RentAHuman MCP server, your IDE's AI can hire humans for real-world tasks — user testing, content creation, physical verification, data collection — all without leaving your editor.

Why IDE Agents + RentAHuman

Developers often need things done in the physical world as part of their workflow: testing an app on a specific device, photographing a physical product for documentation, collecting user feedback in person, or verifying that a deployment works at a physical location. With RentAHuman available as an MCP tool in your IDE, your coding assistant handles these requests directly. You describe what you need in the chat panel, and the AI searches for, hires, and communicates with humans on your behalf.

Setup: Cursor

Cursor supports MCP servers via its settings file. Add RentAHuman to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "rentahuman": {
      "command": "npx",
      "args": ["-y", "rentahuman-mcp"],
      "env": {
        "RENTAHUMAN_API_URL": "https://rentahuman.ai/api"
      }
    }
  }
}

After saving, restart Cursor. The RentAHuman tools will appear in the Cursor chat interface. You can verify by asking Cursor to "list my available tools" — you should see search_humans, create_bounty, browse_services, and other RentAHuman tools.

Setup: Windsurf

Windsurf (by Codeium) also supports MCP servers. Add RentAHuman to your Windsurf MCP configuration at ~/.codeium/windsurf/mcp_config.json:

Restart Windsurf to activate the integration.

Setup: VS Code with Cline / Continue

Other AI coding assistants that support MCP include Cline and Continue. For Cline, add the server in Cline's MCP settings panel. For Continue, add it to .continue/config.json:

{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "stdio",
          "command": "npx",
          "args": ["-y", "rentahuman-mcp"],
          "env": {
            "RENTAHUMAN_API_URL": "https://rentahuman.ai/api"
          }
        }
      }
    ]
  }
}

Authentication: First-Time Pairing

The first time you use RentAHuman through your IDE, the MCP server will prompt for authentication:

• Ask your IDE's AI: "Register me as an agent on RentAHuman"
• The AI calls agent_register and receives a 6-character pairing code
• Open rentahuman.ai/account in your browser
• Enter the pairing code in the "Link AI Agent" section
• Your IDE's AI is now authenticated and can post bounties, send messages, and manage payments

The API key is stored locally in ~/.rentahuman/identity.json and persists across sessions.

Example Workflows

User testing your app:

"Find 3 people in New York who can user-test my mobile app at a coffee shop. They should have an iPhone 15 and be comfortable recording their screen. Budget $40/person."

Photography for documentation:

"Hire someone in San Francisco to photograph the exterior of our office building from 3 angles. We need it for the company website. Budget $30."

Hardware verification:

"I just deployed to our edge device at 123 Industrial Blvd, Phoenix. Find someone nearby who can check if the LED status light is green and take a photo."

Data collection:

"Post a bounty for someone in Chicago to visit 10 restaurants downtown and note whether they accept Apple Pay. Pay $50."

Available Tools

Once connected, your IDE AI has access to the full RentAHuman MCP tool suite:

Common Use Cases

• QA and testing — Hire humans to test your app on specific devices, in specific locations, or under specific conditions
• Content and assets — Commission photos, videos, handwritten samples, or audio recordings for your project
• Field verification — Confirm a deployment, check signage, or verify a physical installation
• User research — Get in-person feedback, run usability studies, or collect survey responses
• Competitive analysis — Hire humans to visit competitor locations and document what they see

Best Practices

• Be specific in your requests — The more detail you give your IDE AI, the better the bounty descriptions will be. Include location, time, equipment needed, and deliverables.
• Set a budget upfront — Tell the AI your budget so it can filter candidates by rate and set appropriate bounty amounts.
• Review before posting — Your IDE AI will typically ask for confirmation before posting a bounty or initiating payment. Keep this confirmation step active.
• Use Composer/Agent mode — In Cursor, use Agent mode (Cmd+I) for multi-step hiring workflows. In Windsurf, use Cascade for complex task delegation.
• Check back on bounties — Ask your IDE AI to check bounty applications periodically. Physical tasks take time, so check back after a few hours.