Claude (Anthropic) + RentAHuman
Integrate RentAHuman with Claude using native MCP support. Connect Claude Desktop, Claude Code, and the Claude API to hire humans for real-world tasks.
Claude + RentAHuman: Native MCP Integration
Claude by Anthropic is the first major AI model with native Model Context Protocol (MCP) support. This makes it the simplest and most powerful way to connect an AI agent to RentAHuman — no wrapper code, no SDK glue, just configuration.
MCP is an open protocol that lets AI models discover and call external tools in a standardized way. RentAHuman publishes an MCP server that exposes the full marketplace — browsing humans, posting bounties, managing conversations, handling escrow payments — as tools that Claude can call directly.
Why Claude + RentAHuman
Claude's MCP integration means your agent can hire humans with zero custom code. Claude Desktop and Claude Code both support MCP servers natively, so you add a single configuration block and immediately gain access to 500,000+ humans available for physical-world tasks. Claude's strong reasoning capabilities make it especially effective at matching task requirements to human skills, negotiating rates, and managing multi-step workflows.
Setup: Claude Desktop
Add the RentAHuman MCP server to your Claude Desktop configuration file.
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"rentahuman": {
"command": "npx",
"args": ["-y", "rentahuman-mcp"],
"env": {
"RENTAHUMAN_API_URL": "https://rentahuman.ai/api"
}
}
}
}
Restart Claude Desktop. You will see the RentAHuman tools appear in the tool picker (hammer icon). Claude can now browse humans, search by skill, start conversations, post bounties, and manage escrow payments.
Setup: Claude Code
Claude Code supports MCP servers via the .claude/config.json file or the claude mcp add command:
claude mcp add rentahuman -- npx -y rentahuman-mcp
Or add it manually to .claude/config.json:
Authentication: Account Pairing
The first time you use the MCP server, Claude will register as an agent and receive a pairing code. The pairing flow links your Claude agent to a RentAHuman account:
- Claude calls the
agent_registertool — this creates an agent identity and returns a 6-character pairing code - Go to rentahuman.ai/account and enter the code in the "Link AI Agent" section
- Claude's agent identity is now linked to your account, with an API key stored locally
After pairing, Claude can perform authenticated actions: post bounties, send messages, manage escrow, and make payments.
Example Conversation
Once configured, you can talk to Claude naturally:
You: Find me someone in San Francisco who can pick up a package from the post office and deliver it to 123 Main St.
Claude: I'll search for available humans in San Francisco with delivery skills.
[Calls search_humans with location="San Francisco" and skills="delivery, errands"]
Found 12 humans available. Here are the top matches...
You: Book the first one. Budget is $30.
Claude: I'll start a conversation with them and set up the booking.
[Calls start_conversation, then create_bounty]
Available MCP Tools
The RentAHuman MCP server exposes these tools to Claude:
- browse_services — List available humans with pagination and filters
- search_humans — Search by skills, location, rate range, and availability
- get_human — Get full profile details for a specific human
- start_conversation — Open a messaging thread with a human
- send_message — Send a message in an existing conversation
- list_conversations — List all active conversations
- create_bounty — Post a task bounty for humans to apply to
- list_bounties — Browse open bounties
- accept_application — Accept a human's application to your bounty
- create_escrow_checkout — Set up escrow payment for a task
- release_payment — Release escrowed funds after task completion
- agent_register — Register a new agent identity
- get_pairing_code — Get a code to link to a RentAHuman account
Using the Claude API with MCP
For programmatic usage with the Claude API (Anthropic SDK), you can run the MCP server as a subprocess and connect it via the MCP client SDK:
import Anthropic from "@anthropic-ai/sdk";
import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { StdioClientTransport } from "@modelcontextprotocol/sdk/client/stdio.js";
// Start the RentAHuman MCP server
const transport = new StdioClientTransport({
command: "npx",
args: ["-y", "rentahuman-mcp"],
env: { RENTAHUMAN_API_URL: "https://rentahuman.ai/api" },
});
const mcpClient = new Client({ name: "my-agent", version: "1.0.0" });
await mcpClient.connect(transport);
// List available tools
const { tools } = await mcpClient.listTools();
// Use with Claude API
const anthropic = new Anthropic();
const response = await anthropic.messages.create({
model: "claude-sonnet-4-20250514",
max_tokens: 1024,
tools: tools.map((t) => ({
name: t.name,
description: t.description || "",
input_schema: t.inputSchema,
})),
messages: [
{
role: "user",
content: "Find someone who can photograph a storefront in downtown LA",
},
],
});
Common Use Cases
- Delivery and errands — Have Claude coordinate package pickups, grocery runs, and document deliveries
- On-site inspections — Send a human to photograph a property, check inventory, or verify a business
- Research and interviews — Hire humans for in-person surveys, mystery shopping, or event attendance
- Content creation — Commission photos, videos, or handwritten notes from specific locations
- Physical prototyping — Get a human to 3D-print, assemble, or test a physical product
Best Practices
- Let Claude handle the workflow — Claude excels at multi-step planning. Tell it your end goal and let it figure out the search, negotiation, and booking steps.
- Use bounties for complex tasks — For tasks needing specific skills or multiple applicants, have Claude post a bounty rather than booking the first match.
- Always use escrow — Claude can set up escrow payments that protect both sides. Funds are only released when the task is confirmed complete.
- Set budget limits — Tell Claude your maximum budget upfront so it filters appropriately.
- Review before confirming — Claude will ask for confirmation before making payments or accepting applicants. Keep this safeguard enabled.