Set up the MCP server

Connect any MCP client to https://heyalbert-mcp-30244774865.us-east1.run.app/mcp: Claude and ChatGPT sign in with OAuth using your HeyAlbert API key, and your own agents send it as a bearer token.

Last updated October 25, 2026

Server details

Claude

claude.ai custom connector (OAuth sign-in)

1. In claude.ai open Settings → Connectors → Add custom connector. On Team and Enterprise plans an owner may need to add it under the admin connector settings first.
2. Name it HeyAlbert and enter the server URL https://heyalbert-mcp-30244774865.us-east1.run.app/mcp.
3. Click Connect. The HeyAlbert sign-in page opens. Paste your API key and confirm.
4. In a new chat, turn HeyAlbert on in the tools menu and ask, for example, “Find founders in Austin with a liquidity event in the last three years.”

The connection uses OAuth 2.1 with PKCE. Claude stores a HeyAlbert token, not your key. Access tokens last 7 days and refresh automatically; they stop working if your key is revoked.

Claude Desktop with your key (config file)

Use this if you prefer a local config. It needs Node.js 18 or later. In Claude Desktop open Settings → Developer → Edit Config and add:

{
  "mcpServers": {
    "heyalbert": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://heyalbert-mcp-30244774865.us-east1.run.app/mcp",
               "--header", "Authorization:${AUTH_HEADER}"],
      "env": { "AUTH_HEADER": "Bearer YOUR_API_KEY" }
    }
  }
}

Quit and reopen Claude Desktop. Keep Authorization:${AUTH_HEADER} exactly as written, with no space after the colon.

ChatGPT Early access

1. In ChatGPT settings, turn on developer mode for apps and connectors (on business plans a workspace admin may need to allow custom connectors). Menu names vary by plan.
2. Create a custom connector named HeyAlbert with the server URL https://heyalbert-mcp-30244774865.us-east1.run.app/mcp and OAuth authentication.
3. Click Connect and paste your API key on the HeyAlbert sign-in page.
4. Pick HeyAlbert from the tools menu in a chat. Deep research can use the search and fetch tools.

A dedicated ChatGPT app with visual person cards is on the roadmap.

Your own agent

Any MCP client or agent framework can connect over Streamable HTTP with the header Authorization: Bearer YOUR_API_KEY. The server is stateless: a normal MCP initialize is all the handshake it needs.

curl

curl -s https://heyalbert-mcp-30244774865.us-east1.run.app/mcp \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"search_people",
       "arguments":{"metro":"Dallas","role_keywords":["physician"],"limit":5}}}'

Python, with the official MCP SDK

import asyncio
from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client

URL = "https://heyalbert-mcp-30244774865.us-east1.run.app/mcp"
HEADERS = {"Authorization": "Bearer YOUR_API_KEY"}

async def main():
    async with streamablehttp_client(URL, headers=HEADERS) as (read, write, _):
        async with ClientSession(read, write) as session:
            await session.initialize()
            res = await session.call_tool("search_people", {"metro": "Austin", "limit": 3})
            for p in res.structuredContent["results"]:
                print(p["name"], p["title"], p["id"])

asyncio.run(main())

Every tool returns clean JSON in structuredContent plus a short text summary. See the tools reference and limits & errors.