Claude custom connector
Add the remote MCP URL in Claude's connector settings, then enable it in a conversation.
Name: Caboo
Remote MCP server URL: https://getcaboo.com/mcp
Auth: None
Try:
"Find caboo-demo-live and hold the first open consultation slot."
Claude-compatible MCP config
For MCP clients that accept a JSON config, use the remote URL directly.
{
"mcpServers": {
"caboo": {
"url": "https://getcaboo.com/mcp"
}
}
}
OpenAI Agents SDK
Use the Streamable HTTP MCP server when your Python process should manage the connection.
import asyncio
from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp
async def main():
async with MCPServerStreamableHttp(
name="Caboo",
params={"url": "https://getcaboo.com/mcp"},
cache_tools_list=True,
) as caboo:
agent = Agent(
name="Booking assistant",
instructions=(
"Use Caboo to find providers and hold slots. "
"Never collect contact details in chat; give the confirmUrl."
),
mcp_servers=[caboo],
)
result = await Runner.run(
agent,
"Find caboo-demo-live and hold the first open slot.",
)
print(result.final_output)
asyncio.run(main())
OpenAI hosted MCP tool
Use the hosted MCP tool when the Responses API should call the public server directly.
from agents import Agent, HostedMCPTool, Runner
agent = Agent(
name="Booking assistant",
instructions="Use Caboo for local-service booking holds.",
tools=[
HostedMCPTool(
tool_config={
"type": "mcp",
"server_label": "caboo",
"server_url": "https://getcaboo.com/mcp",
"require_approval": "never",
}
)
],
)
result = await Runner.run(
agent,
"Find caboo-demo-live and show me open appointment times.",
)
LangChain / LangGraph
Load Caboo's remote MCP tools with the HTTP transport, then pass them into your agent.
from langchain_mcp_adapters.client import MultiServerMCPClient
from langchain.agents import create_agent
client = MultiServerMCPClient({
"caboo": {
"transport": "http",
"url": "https://getcaboo.com/mcp",
},
})
tools = await client.get_tools()
agent = create_agent("openai:gpt-5.4", tools)
result = await agent.ainvoke({
"messages": "Find caboo-demo-live and hold an open slot."
})
Vercel AI SDK
Use @ai-sdk/mcp to convert remote MCP tools into AI SDK tools.
import { createMCPClient } from "@ai-sdk/mcp";
import { generateText, stepCountIs } from "ai";
const caboo = await createMCPClient({
transport: {
type: "http",
url: "https://getcaboo.com/mcp",
redirect: "error",
},
name: "caboo-demo",
version: "1.0.0",
});
try {
const tools = await caboo.tools();
const result = await generateText({
model: "openai/gpt-5.4",
tools,
stopWhen: stepCountIs(8),
prompt: "Find caboo-demo-live and hold the first open slot.",
});
console.log(result.text);
} finally {
await caboo.close();
}