WATI MCP server for your AI

The Access token on its own covers every modern (v3) operation: contacts, messaging, template sends, broadcasts, channels, chatbots and sales analytics. The API endpoint URL is optional and exists only so your tenant id can be filled in for the legacy operations — creating and deleting templates, WABA and phone-number management, and webhook registration. Add it up front if you expect to touch any of those; otherwise the token alone is fine.

send_session_text is only valid inside the 24-hour customer-service window that opens when a customer messages your business number. Once that window closes, WhatsApp allows approved templates only, so the agent has to switch to send_template_messages or send_template_message_v2. Run list_templates first and let the agent pick an approved template when the window has lapsed.

Yes, but only through the legacy path — create_template_v1 submits a template for approval and delete_template_v1 removes one by WABA id and name, because v3 cannot create templates. That means template creation needs the API endpoint URL saved on the connection. Approval still sits with WhatsApp, so watch the Template submitted, Template approved and Template rejected or paused triggers for the outcome rather than expecting an instant answer.

send_template_messages takes 1 to 10,000 recipients in one call, which is the modern route for a campaign-sized send. The legacy bulk endpoint, send_template_messages_bulk_v1, tops out at 1000. For a one-to-one send there are single-recipient variants — send_template_message_v1 and send_template_message_v2.

The access token is scoped to one WATI account, and the agent can see every channel inside it. list_channels returns the WhatsApp numbers connected to the account, list_phone_numbers_v2 adds channel and WABA detail (v2 beta), and list_business_accounts_v1 lists the WABAs on the tenant — that last one is legacy, so it needs the API endpoint URL. The Channel connected or disconnected and Number quality rating changed triggers tell you when that picture shifts.

Yes. get_messages pulls the history for a conversation, assign_conversation_operator hands a thread to an operator by email, and update_conversation_status moves it to open, resolved or another state. start_chatbot can kick off a bot flow for a contact, and assign_contact_to_teams routes the person to the right team. Pair that with the Customer sent a message trigger and the agent can triage as chats arrive.

No. PopMCP hosts it and gives you a Streamable HTTP MCP URL. There is nothing to deploy, no callback URL to register, and no infrastructure to keep running or patch.

Any client that supports a remote Streamable HTTP MCP server — Claude Desktop, Claude Web, ChatGPT custom connectors, Codex and the Codex CLI, OpenCode and Z.ai's ZCode, as well as editors such as Cursor, Windsurf and Zed. Your client signs in to PopMCP with OAuth; the provider credential never leaves the server.

Loading 60 tool definitions into every conversation burns context before the work starts. PopMCP loads a curated 30-tool set covering the everyday workflows and keeps the rest searchable, so your AI pulls a specialist tool only when a task needs it. You can switch a connection to the full catalog at any time.

A tool is something your AI calls — there are 60 here. A trigger is something PopMCP watches for you: 14 WATI events that can start work on the PopMCP side. Triggers never enter your AI client's context, and none of them ask you to paste a callback URL into WATI.