CdpHub MCP Server已验证已精选

Ask your AI assistant which customer data platforms (CDPs) integrate with your tools, check a whole stack against every CDP, and compare CDP integration catalogs.

CdpHub is an independent directory of CDP integrations. It brings the public integration catalogs of more than 40 CDPs, including Segment, mParticle, Hightouch, Braze, Klaviyo and Adobe Experience Platform, into one place, with every listing linked to the CDP's own documentation. This MCP server lets Claude, ChatGPT, Cursor, VS Code and other MCP clients look things up in it.

• Server address: https://cdphub.com/functions/mcp (Streamable HTTP)
• Sign-in: none. No API key, no account, no session.
• Read-only: six lookup tools. It can't change anything.
• Official MCP Registry name: com.cdphub/cdphub
• Setup page: https://cdphub.com/mcp
• Docs on GitHub: https://github.com/cdphub/cdphub-mcp

Connect

Claude

CdpHub is in Claude's Connectors Directory:

1. Open CdpHub in Claude's Connectors Directory, or search for CdpHub under Customize › Connectors.
2. Click Connect. There's no sign-in.
3. In a chat, click +, then Connectors, and switch on CdpHub.

On Team and Enterprise plans, an owner adds it under Organization settings › Connectors. To add it by hand instead, choose + Add, then Add custom connector, paste the server address and keep No sign-in.

ChatGPT

1. Open Plugins (chatgpt.com/plugins).
2. Click +, then Add custom MCP server.
3. Name it CdpHub and paste the server address under Connection.
4. Choose no authentication, confirm the warning and click Create as a plugin.
5. Install the plugin. In a new chat, type @ and pick CdpHub.

Whether you can add a custom MCP server depends on your ChatGPT plan and workspace settings.

Claude Code

claude mcp add --transport http cdphub https://cdphub.com/functions/mcp

Add --scope user to have it in every project. claude mcp list shows whether it's connected.

Cursor

Add CdpHub to ~/.cursor/mcp.json for every project, or to .cursor/mcp.json in one project:

{
  "mcpServers": {
    "cdphub": {
      "url": "https://cdphub.com/functions/mcp"
    }
  }
}

If the file already lists servers, add the cdphub entry inside mcpServers.

VS Code (GitHub Copilot)

Run MCP: Add Server from the Command Palette, choose HTTP, paste the server address and name it cdphub. Or add it to .vscode/mcp.json:

{
  "servers": {
    "cdphub": {
      "type": "http",
      "url": "https://cdphub.com/functions/mcp"
    }
  }
}

Other apps

Any app that supports remote MCP servers over Streamable HTTP can connect with the server address alone. Opened in a browser, the address takes you to the setup page.

Tools

Tool	What it answers	Try asking
get_tool	One tool: which CDPs list an integration with it, which way the data flows (sends, reads or both), how it runs (server-side, client-side or in the warehouse), with links to each CDP's documentation	"Which CDPs integrate with Snowflake, and which way does the data flow?"
check_stack	Two to ten tools together: the CDPs that list all of them, grouped by how they connect to each, and the ones that miss just one	"Which CDPs list Shopify, Klaviyo, Google Ads and BigQuery together?"
compare_cdps	Two CDPs' catalogs, tool by tool: the tools both list, the tools only one lists, and each side's largest categories of those	"Compare Segment's and mParticle's integration catalogs."
get_cdp	One CDP's catalog: size and rank, directions, largest categories, how it connects to the main data warehouses, and the CDPs with the most tools in common	"What does Braze's integration catalog cover, and how does it connect to the main warehouses?"
list_cdps	Every CDP in the directory by catalog size, with the directory's totals	"Which CDPs have the largest integration catalogs?"
search_directory	Searches by keyword, CDP, category, direction or how the integration runs, with exact totals and the first results	"Search the CDP directory for reverse ETL integrations."

Parameters:

• get_tool: name (required), for example "Snowflake".
• check_stack: tools (required), two to ten tool names, for example ["Shopify", "Klaviyo", "Snowflake"].
• compare_cdps: a and b (both required), for example "Segment" and "mParticle".
• get_cdp: name (required), for example "Braze".
• list_cdps: none.
• search_directory: all optional. query (words to search for), cdps and categories (lists of names), directions (out, into, both, unknown), runs (server, client, warehouse, none), group (tool or listing), sort (relevance, coverage, az, added, recommended).

All six tools are annotated read-only, not destructive, idempotent and closed-world.

Reading the answers

• Each CDP is counted once per tool, by the same rules for every CDP.
• A CDP with no listing for a tool doesn't document an integration with it. That isn't proof the two can't be connected.
• A tool lookup details up to 50 listings; the tool's page on cdphub.com has them all.
• Every answer links to the matching page on cdphub.com and carries the date of its data.

Terms and privacy

• Free for answering questions, with a link back to cdphub.com. For bulk data, change alerts or commercial use of the data, email ed@cdphub.com about API access. Terms: https://cdphub.com/terms
• CdpHub records each lookup (the tool and what it was asked) and the name of each app that connects, so it can see which questions people ask. No IP address is stored. Privacy policy: https://cdphub.com/privacy-policy
• Lookups are rate-limited so the server stays quick for everyone.

CdpHub is independent and takes no referral fees or vendor commissions.

Contact

Questions or problems: ed@cdphub.com or https://cdphub.com/contact

© CdpHub LLC