Hello Clever Developer Docs

Hello Clever hosts a Model Context Protocol (MCP) server for this documentation site. Connect it to your AI coding assistant and it can search and read every published page while it answers you, API reference and OpenAPI specs included. Why it helps: without the MCP server, your assistant is working from memory. It might invent an endpoint that does not exist, reach for a field name from an older version, or mix Hello Clever up with another payment provider. Once you connect it, it checks the current page before it answers, so you spend less time correcting it. Setup takes a minute or two, and there is nothing to install. The server is free, public, read-only, and needs no API key.

So you know which server this is: it reads our published documentation and nothing else. It is not an API into your Hello Clever account, so it cannot see your merchant data, balances, or transactions. If you want answers about your own payment data, Clever AI in the Merchant Portal is the tool for that.

Server details

Everything you need to set it up:

Property	Value
Server URL	`https://docs.helloclever.co/mcp`
Transport	HTTP (streamable)
Authentication	None. The server is public.
Access	Read-only
Discovery	`https://docs.helloclever.co/.well-known/mcp`
Server card	`https://docs.helloclever.co/.well-known/mcp/server-card.json`

The server introduces itself as Hello Clever. Use the URL exactly as it appears above: there is no separate host, port, or npx package to install.

What your AI app gets

Tools

Three of them, and your assistant chooses between them on its own.

Search

Searches the whole site and hands back matching snippets with page titles and links. Great for broad or conceptual questions like “how do I authenticate”.search_hello_clever

Read the docs

Reads the docs as a virtual read-only filesystem using ls, tree, rg, cat, head, and jq. Handy for exact keyword matches, browsing the structure, or pulling up a full page.query_docs_filesystem_hello_clever

Submit feedback

Sends a documentation problem back to our docs team: a page that is wrong, outdated, incomplete, or has an example that does not work.submit_feedback

A typical run searches first, then reads the most promising pages in full.

The name of the second tool sounds alarming, so to be clear: query_docs_filesystem_hello_clever never touches your machine. It runs against an in-memory copy of the published documentation on our side, with no network access, no write access, and no visibility into your local files.

Resources

You also get a helloclever skill file, exposed as an MCP resource. If the tools give your assistant access to the content, the skill tells it how to use that content: which flows apply to payins and payouts, how to configure webhooks, and how to test in sandbox. Clients that support MCP resources pick it up automatically once you connect, so there is nothing extra to install.

What it covers

Everything in our published navigation, in every language the site ships:

Getting Started, Platform Overview, payment concepts, and Clever Concepts
Merchant Portal and Clever AI guides
Platform integrations such as Shopify, WooCommerce, Magento, and Xero
The full API Reference, including the v1, v2, Payment Gateway 3, Card, and Cashback OpenAPI specs
Security and compliance pages

Connect from the docs site

If you would rather not touch a config file, start here.

Open the menu

At the top right of any page on this site, select the arrow next to the Copy page button.

Pick your option

Copy MCP Server puts https://docs.helloclever.co/mcp on your clipboard, ready to paste into your client. Connect to Cursor installs it in Cursor for you in one click.

The same menu is handy even without the MCP server. Open in ChatGPT and Open in Claude start a conversation about the page you are reading, and Copy page and View as Markdown give you that page as plain Markdown to paste anywhere you like.

Connect your AI app

Pick your tool below.

One command in your terminal:

claude mcp add --transport http hello-clever https://docs.helloclever.co/mcp

Add --scope project if you would like to commit the server to your repository’s .mcp.json and share it with your team, or --scope user to have it available in every project on your machine.To confirm, run /mcp inside Claude Code. You should see the server listed as connected.

Open connector settings

Head to the Connectors page in your Claude settings.

Add a custom connector

Select Add custom connector, then enter:

Name: Hello Clever Docs
URL: https://docs.helloclever.co/mcp

Save it

Select Add.

Use it in a chat

Select the attachments button (the plus icon) in any conversation, then select Hello Clever Docs.

The quickest way is Connect to Cursor, in the menu behind the Copy page button at the top of this page. If you prefer to set it up yourself:

Open MCP settings

Press Cmd + Shift + P (Ctrl + Shift + P on Windows), search for Open MCP settings, then select Add custom MCP. This opens mcp.json.

Add the server

mcp.json

{
  "mcpServers": {
    "hello-clever": {
      "url": "https://docs.helloclever.co/mcp"
    }
  }
}

Create .vscode/mcp.json in your project:

.vscode/mcp.json

{
  "servers": {
    "hello-clever": {
      "type": "http",
      "url": "https://docs.helloclever.co/mcp"
    }
  }
}

Commit the file and everyone working in the repository gets the server too.

Add the server to ~/.codex/config.toml:

~/.codex/config.toml

[mcp_servers.hello-clever]
url = "https://docs.helloclever.co/mcp"

Most other agents accept the URL through the add-mcp helper, which writes the right config for whichever client you pick:

npx add-mcp https://docs.helloclever.co/mcp

Not seeing your client? Add it by hand wherever it keeps remote MCP servers. All it needs is the HTTP transport and the URL https://docs.helloclever.co/mcp. No headers, no credentials.

Check that it worked

Ask your assistant something our docs answer precisely and the open web does not. Any of these make a good test:

Using the Hello Clever docs MCP, what fields does a PayTo agreement need?
Search the Hello Clever docs for the difference between a Payments Account and a Treasury Account.
Read the v2 Payin endpoint from the Hello Clever docs and write me a curl request.

If the answer comes back with page links to docs.helloclever.co, you are connected. If it stays vague and cites nothing, it is still working from memory: check that the server shows as connected in your client, then try again.

Call the server from your own code

Building your own integration? Any MCP client library can talk to the server. To try it from a terminal first, send a JSON-RPC initialize call:

curl -X POST https://docs.helloclever.co/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "initialize",
    "params": {
      "protocolVersion": "2025-06-18",
      "capabilities": {},
      "clientInfo": {"name": "my-integration", "version": "1.0.0"}
    }
  }'

curl -X POST https://docs.helloclever.co/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc": "2.0", "id": 2, "method": "tools/list"}'

curl -X POST https://docs.helloclever.co/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{
    "jsonrpc": "2.0",
    "id": 3,
    "method": "tools/call",
    "params": {
      "name": "search_hello_clever",
      "arguments": {"query": "PayTo agreement"}
    }
  }'

Two things worth knowing. Responses come back as server-sent events, so keep text/event-stream in your Accept header. And the search tool takes an optional language parameter, for example "language": "ja", if you want results from the Japanese pages. It defaults to en.

Rate limits

These limits keep the server available for everyone. Normal interactive use will not come close to them.

Scope	Limit
Per user (IP address)	5,000 requests per hour
Search, per site	10,000 requests per hour
Query docs filesystem, per site	10,000 requests per hour

If you do go over, the server returns 429 Too Many Requests. Back off and retry: limits reset on a rolling hourly window, so you will not be locked out for long.

If something goes wrong

The server will not connect

Check the URL is exactly https://docs.helloclever.co/mcp, with no trailing slash and nothing after it. Then check your client is set to the HTTP transport rather than stdio or SSE. One thing that is not a problem: a 405 response if you open that URL in a browser. The endpoint only answers POST, so that is expected.

It connects, but never seems to use the tools

Some clients only reach for MCP tools when the question clearly calls for them. Try naming the source in your prompt, for example “search the Hello Clever docs for…”. In Claude web and desktop, also check the connector is selected through the attachments button for that conversation.

I am getting answers from the old documentation

The server indexes this site only. Anything on legacy-docs.helloclever.co is outside the index, so answers citing it are coming from your model’s training data or a web search rather than from us.

A page my assistant quoted is wrong

Please tell us. Ask your assistant to use the submit_feedback tool with the page path and a description of the problem, and it reaches our docs team directly.

Security and privacy

Connecting the server is low risk. Here are the specifics:

The server is read-only. The only tool that sends anything back to us is submit_feedback, and all it sends is the page path and the feedback text your assistant supplies.
It serves published documentation. None of it is account-specific, so connecting it exposes none of your data to us.
Your prompts stay with your AI provider. Only the search queries and file commands your assistant issues reach our documentation server.
Adding the server to a shared config file such as .mcp.json or .vscode/mcp.json is safe to commit. There are no secrets in it.

For how we handle data more broadly, see How we handle data and API security.

Where to go next

Choosing an integration: pick your integration path before you start building
Integration keys: the credentials your code needs, which the MCP server never handles
API overview and Authentication: the reference pages your assistant will read most often
Clever AI tutorials: the AI built into the Merchant Portal, which does read your payment data

Integration Basics

Platform Integrations

API Use Cases

Clever AI Connectors

AI Tools

Security & Compliance

Connect Your AI App to the Docs MCP Server

Server details

What your AI app gets

Tools

Search

Read the docs

Submit feedback

Resources

What it covers

Connect from the docs site

Connect your AI app

Check that it worked

Call the server from your own code

Rate limits

If something goes wrong

Security and privacy

Where to go next

​Server details

​What your AI app gets

​Tools

Search

Read the docs

Submit feedback

​Resources

​What it covers

​Connect from the docs site

​Connect your AI app

​Check that it worked

​Call the server from your own code

​Rate limits

​If something goes wrong

​Security and privacy

​Where to go next

Server details

What your AI app gets

Tools

Resources

What it covers

Connect from the docs site

Connect your AI app

Check that it worked

Call the server from your own code

Rate limits

If something goes wrong

Security and privacy

Where to go next