Docs menuGetting started
MCP documentationBahasa Indonesia

Klik Flow MCP: your AI works with your business data.

With MCP, AI apps such as Claude, ChatGPT, Codex and Cursor can read chats, contacts, pipeline, reports and other Klik Flow modules, prepare replies and tidy up conversation data, within your role's permissions. One address for every app: https://mcp.klikflow.id/mcp.

One address for every AI appNormal Klik Flow sign-in, no keys to pasteWithin your role's permissionsDisconnect at any time

What is Klik Flow MCP?

MCP (Model Context Protocol) is an open standard for connecting AI apps to other systems. Klik Flow runs one official MCP server at https://mcp.klikflow.id/mcp. After you sign in and approve its permissions, an AI app can read business data within your role, for example summarise today's chats, and then prepare replies that staff approve.

Requirements

The business's package includes the AI Connection feature (Koneksi AI), and your role has the AI Connection › Connect permission (Koneksi AI › Hubungkan; owners always have it). One connection covers one business; for another business, connect again from the AI app. The Klik Flow app and its consent page are in Indonesian, so these guides give the Indonesian names next to the English ones.

  • The owner sees and ends every connection in Settings › AI Connection (Pengaturan › Koneksi AI).
  • Agents without a browser (servers, n8n) use an access key (Kunci akses) from the same page.

What the AI can read

Chats and messages, contacts, pipeline, tickets, social comments, follow-ups and upsells, broadcasts, landing pages, Bio Link, Smart Link, content, media, the store and orders, reports, ad signals, automations, AI Agent, quick replies and Prospect Finder (Cari Prospek). Modules that are not ticked, or that your role cannot see, do not appear at all.

  • Klik Flow MCP is in beta. New tools are added step by step and listed in the Reference.

What the AI can change

A chat's status, assignee, labels and AI mode; team notes; contacts, tags and reminders; pipeline cards, including won or lost; tickets; chat replies, chats to new numbers, comment replies and comment DMs, and follow-ups. For marketing: landing page drafts (text, images, order, look) and publishing; Bio Link blocks and publishing; Smart Link; content drafts, captions from Klik Flow AI, schedules and content tasks; broadcast drafts with a recipient preview and then a send request; and image uploads to Media. Every action needs the permission ticked at consent, the same role permission as its button in the app, and the package feature.

  • By default, replies to customers become drafts in the Inbox that staff send (the Draft policy).
  • By default, publishing landing pages and Bio Link, changing a live Smart Link, scheduling content and sending broadcasts wait in the Awaiting approval panel (Menunggu persetujuan) of their module.
  • Deleting data, payments, HTML code, pixels and settings are never open to the AI.

Official address

Use only https://mcp.klikflow.id/mcp. Klik Flow does not publish any npm package, plugin or skill called "Klik Flow MCP"; do not install one from another source that claims to be official.

Steps

  1. Add the server In your AI app, add an MCP server with the address https://mcp.klikflow.id/mcp. Each app has its own guide in this documentation.
  2. Sign in and approve The browser opens the Klik Flow sign-in, then the consent page: pick the business, the permissions, the send policy and how long the connection lasts.
  3. Start asking Try "Summarise today's chats" or a ready-made prompt such as laporan-mingguan (weekly report).

Frequently asked questions

Can the AI reply to customers?

Yes. By default, AI replies become drafts in the Inbox, marked with the AI app's name and the person who asked for them; staff send, edit or discard them. If the package includes the AI send feature, you can choose Confirm in the AI app, and the owner can choose Direct with limits.

Does it use AI credits?

Reading data over MCP does not use Klik Flow AI credits, because the AI is your own. Tools that run Klik Flow's own AI, such as captions from Klik Flow AI, use AI credits as usual.

My AI app only supports local (stdio) servers.

Use the mcp-remote bridge, version 0.1.16 or later: npx -y mcp-remote@latest https://mcp.klikflow.id/mcp. Older versions have a serious security flaw.

Connect Klik Flow to your AI app.

Mulai dengan paket Free. Ajak tim setelah ruang kerja siap.