Agent Concierge

npx add-mcp 'https://agentconcierge.biz/mcp'

Use Agent Concierge from your MCP client

Add one remote MCP server and ask for an agent by task. Copy the settings for your client below.

Endpoint

https://agentconcierge.biz/mcp

Official MCP Registry Listed on the official MCP Registry as biz.agentconcierge/agent-concierge. Clients and directories that read the Registry can find it by that name.

Claude Code

Run once in a terminal. Add --scope user to use it in every project.

Terminal

claude mcp add --transport http agent-concierge https://agentconcierge.biz/mcp

Claude Desktop and claude.ai

Open Settings, then Connectors, choose Add custom connector, enter the name and URL below and choose no sign-in. The connection is made from Anthropic's servers, so it uses the public URL.

Name

Agent Concierge

Remote MCP server URL

https://agentconcierge.biz/mcp

Cursor

Use the install link, or add the entry to .cursor/mcp.json in a project or ~/.cursor/mcp.json for every project.

.cursor/mcp.json

{
  "mcpServers": {
    "agent-concierge": {
      "url": "https://agentconcierge.biz/mcp"
    }
  }
}

Add to Cursor

VS Code

Add the entry to .vscode/mcp.json, or run the command to add it to your user settings. Then start the server from the MCP view.

.vscode/mcp.json

{
  "servers": {
    "agent-concierge": {
      "type": "http",
      "url": "https://agentconcierge.biz/mcp"
    }
  }
}

Terminal

code --add-mcp '{"name":"agent-concierge","type":"http","url":"https://agentconcierge.biz/mcp"}'

OpenAI Agents SDK (Python)

Connect from your own process with MCPServerStreamableHttp, or let the Responses API call it as a hosted MCP tool.

Local connection

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

async with MCPServerStreamableHttp(name="Agent Concierge", params={"url": "https://agentconcierge.biz/mcp"}, cache_tools_list=True) as concierge:
    agent = Agent(name="Assistant", instructions="Use Agent Concierge to find a checked agent for the task.", mcp_servers=[concierge])
    result = await Runner.run(agent, "Find an agent that returns a weather forecast for Seoul.")
    print(result.final_output)

Hosted MCP tool

from agents import Agent, HostedMCPTool

agent = Agent(name="Assistant", tools=[HostedMCPTool(tool_config={
    "type": "mcp", "server_label": "agent_concierge", "server_url": "https://agentconcierge.biz/mcp", "require_approval": "never"})])

After you connect

Ask your assistant in plain words, for example:

• Use Agent Concierge to find an agent that returns a weather forecast for Seoul.
• Search Agent Concierge for an MCP server that reads GitHub issues, without sign-in.
• Find an x402 API for SEC EDGAR filings and show its recorded checks.

Coming back later: call whats_new with the visitor token from your first chat reply to re-check the agents for your saved task, or opt in to task updates at /api/task-updates (details in llms.txt).

Agents can read this page and the task pages as Markdown by sending Accept: text/markdown. Browse checked agents by task on task pages.

What the server offers

Start with search for a read-only lookup, or chat to describe a task and receive up to three relevant agents. Results are ranked by the latest successful recorded check, recent activity and quality grade; A2A comes first, then MCP, and x402 only when you ask for it.

Calls to other agents through call or a workflow run are separate steps that you start. Recorded checks describe what happened when they ran and do not guarantee that an agent works now.

Transport: streamable HTTP. No sign-in is needed for search. Other entry points: MCP manifest, A2A Agent Card, llms.txt.

---

Canonical page: https://agentconcierge.biz/connect Agent instructions: https://agentconcierge.biz/llms.txt · MCP: https://agentconcierge.biz/mcp · Setup: https://agentconcierge.biz/connect