Install Agentropolis in your agent

Agentropolis is a remote MCP server (Streamable HTTP) that searches MCP servers, x402/MPP pay-per-call APIs and A2A agents, with payment-verified trust. Free, no signup, no API key.

Add to Cursor Install in VS Code Claude Code More clients & SDKsServer URL: https://agentropolis.io/api/mcp

Clients and frameworks

Claude Code

claude mcp add --transport http agentropolis https://agentropolis.io/api/mcp

Any MCP client that accepts a remote URL (mcp.json: Cursor, Windsurf and others)

{
  "mcpServers": {
    "agentropolis": {
      "url": "https://agentropolis.io/api/mcp"
    }
  }
}

VS Code (mcp.json)

{
  "servers": {
    "agentropolis": {
      "type": "http",
      "url": "https://agentropolis.io/api/mcp"
    }
  }
}

Vercel AI SDK (@ai-sdk/mcp)

import { createMCPClient } from '@ai-sdk/mcp';
import { generateText } from 'ai';

const mcp = await createMCPClient({ transport: { type: 'http', url: 'https://agentropolis.io/api/mcp' } });
try {
  const { text } = await generateText({
    model: 'openai/gpt-5-mini', // any AI SDK model
    tools: await mcp.tools(),
    prompt: 'Find an x402 web search API under $0.01 per call with verified buyers.',
  });
  console.log(text);
} finally {
  await mcp.close();
}

LangChain (langchain-mcp-adapters)

from langchain_mcp_adapters.client import MultiServerMCPClient
from langchain.agents import create_agent

client = MultiServerMCPClient({
    "agentropolis": {"transport": "streamable_http", "url": "https://agentropolis.io/api/mcp"}
})
tools = await client.get_tools()
agent = create_agent("openai:gpt-5-mini", tools)
result = await agent.ainvoke({"messages": [{"role": "user", "content": "Find an MCP server for Postgres"}]})

CrewAI

from crewai import Agent
from crewai.mcp import MCPServerHTTP

scout = Agent(
    role="Tool scout",
    goal="Find the right paid API or MCP server for a task",
    backstory="Checks price and payment-verified trust before recommending a tool.",
    mcps=[MCPServerHTTP(url="https://agentropolis.io/api/mcp", cache_tools_list=True)],
)

OpenAI Agents SDK (Python)

from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

async with MCPServerStreamableHttp(name="agentropolis", params={"url": "https://agentropolis.io/api/mcp"}) as server:
    agent = Agent(name="Tool scout", instructions="Find tools for the user's task.", mcp_servers=[server])
    result = await Runner.run(agent, "Find a transcription API with speaker labels")
    print(result.final_output)

Plain HTTP (no MCP)

curl "https://agentropolis.io/api/v1/search?q=web+search&rail=x402&max_price_usd=0.01"

No auth, no signup. All tools are read-only except call_kit, a non-custodial x402 pass-through that never pays on its own: you sign any payment with your own wallet. Links that don't open your app? Use the web versions: Cursor · VS Code.

Full examples: search and pay with x402

Runnable, tested examples (MIT) for Vercel AI SDK, LangChain, CrewAI, OpenAI Agents SDK and Coinbase AgentKit: connect to the MCP server, search, and pay a kit via x402 with your own wallet (key from an env var). github.com/seenly721/agentropolis-examples

What your agent gets

search_kits: find tools for a task, filtered by kind, category, rail, price and trust.
get_kit / get_kit_manifest: trust evidence and price, or the invoke contract (endpoints, schemas, payment offers).
call_kit: optional non-custodial x402 pass-through. It never pays by itself; you sign any payment with your own wallet.
list_categories, list_sources, get_trust_method, get_market_snapshot, get_bazaar_link, get_roadmap.

Other surfaces: A2A agent card /.well-known/agent-card.json · REST /api/v1/search · OpenAPI · llms.txt