How to Connect Your AI Agent to Real Humans via MCP

The Model Context Protocol (MCP) has changed how AI agents discover and use tools. Instead of hardcoding API integrations, agents can dynamically discover what's available and use it immediately. SiliconBridge is built MCP-native — every one of our 22 human-powered services is available as an MCP tool.

What Is MCP?

MCP is a protocol that lets AI models connect to external tools and data sources through a standardized interface. Think of it like USB for AI — plug in a server, and the agent automatically knows what tools are available, what parameters they accept, and how to call them. Claude, Cursor, Windsurf, and many other AI tools support MCP natively.

Adding SiliconBridge to Your Agent

There are two ways to connect: Streamable HTTP (recommended) and SSE (legacy). Both give your agent access to all 22 services.

Option 1: Streamable HTTP (Recommended)

Add this to your MCP configuration (e.g., claude_desktop_config.json for Claude Desktop, or .cursor/mcp.json for Cursor):

{
  "mcpServers": {
    "siliconbridge": {
      "url": "https://siliconbridge.xyz/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Option 2: SSE (Legacy)

{
  "mcpServers": {
    "siliconbridge": {
      "url": "https://siliconbridge.xyz/sse",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

What Your Agent Gets

Once connected, your agent can call any of these services as MCP tools. The agent sees them as native capabilities — it doesn't need to know that a human is involved behind the scenes.

For $1 each, your agent gets digital bridge services: web browsing through bot-protected sites, sending messages via Gmail/Telegram/Slack, creating calendar events, storing persistent memory, and controlling smart home devices. Verification services like CAPTCHA solving, OTP relay, and phone verification run $1-$1.50. For bigger tasks like domain registration, wallet setup, or robot rental, the agent requests a quote with a $1 deposit.

Auto-Discovery

Agents that support dynamic tool discovery can also use our REST discovery endpoints:

# Zero-config onboarding — returns everything in one call
GET https://siliconbridge.xyz/api/discover

# Service catalog with pricing
GET https://siliconbridge.xyz/api/services

# Live platform status
GET https://siliconbridge.xyz/api/status

# LLM-readable description
GET https://siliconbridge.xyz/.well-known/llms.txt

# Agent-to-Agent protocol
GET https://siliconbridge.xyz/.well-known/agents.json

Real Example: Agent Solving a CAPTCHA

Here's what happens when your agent encounters a CAPTCHA while browsing:

# Agent hits a CAPTCHA → calls SiliconBridge MCP tool
Tool: siliconbridge_submit_task
Input: {
  "service_type": "captcha_solve",
  "payload": {"image_url": "https://example.com/captcha.png"}
}

# Human solves it in ~30 seconds
# Agent receives:
{
  "task_id": "task_abc123",
  "status": "completed",
  "result": {"solution": "xK9mP2"}
}

# Agent continues with the solution

Getting Started

Get an API key in 30 seconds — no email required. Use a wallet address or Nostr npub:

curl -X POST https://siliconbridge.xyz/api/signup/wallet \
  -H "Content-Type: application/json" \
  -d '{"wallet_address": "YOUR_WALLET"}'

Then add the MCP config above and your agent is ready. $10 free credits on signup, pay-per-task after that. No subscriptions.