Skip to main content
These are the 43 abilities your assistant gains once ChatSyncs is connected. You never call them by name — you ask in plain English and the assistant picks. They’re listed so you know what’s possible.
Reading the table. Read means it only looks something up and changes nothing. Destructive means it removes or blocks something, so your AI app is told to ask for a firmer confirmation. Everything else writes or sends.

Messaging

send_text_message, send_file and send_interactive_buttons only reach someone who messaged you in the last 24 hours. Outside that window, use an approved template — see Understanding conversations and their types.

Contacts

Labels, fields and sequences

Team

Deals (CRM)

Shop and orders

Help and documentation

These look up ChatSyncs’ own public help — the same guides and video lessons published on help.chatsyncs.com, and the API and MCP docs on developer.chatsyncs.com. They never touch your account or your data, so the assistant can answer “how do I…?” questions accurately instead of guessing.

Account

Limiting what the assistant can do

You can’t get a read-only API key, but you have two other controls:
  • In your AI app. Most apps let you switch individual abilities off or require approval for each one. Claude: Customize → Connectors → ChatSyncs. ChatGPT: the app’s details page in its settings. VS Code: Configure Tools in the chat box. Cline: leave autoApprove empty.
  • A read-only server. ChatSyncs can run a setup that offers only the 20 read abilities above and no way to send or change anything. It’s a separate server rather than an option on mcp.chatsyncs.com; contact ChatSyncs support if you need it.

Frequently asked

Deliberately. The Developer API has no bulk-send endpoint, so a broadcast would mean the assistant messaging people one at a time — slow, and easy to get badly wrong at scale. Use Broadcasting in ChatSyncs, which is built for it.
A few kinds cannot be sent from here, and the assistant tells you which and why before sending anything: carousel templates, templates with an image, video or document header, and templates whose variables are not mapped in ChatSyncs. Send those from ChatSyncs itself. Text templates, with or without variables, work.
The contact’s first name fills itself in. For anything else, such as an order total, the assistant asks you for the value, and it will not send until it has one — otherwise ChatSyncs would deliver a dash in its place. That includes the end of a button’s link: for a Pay Now button that opens https://example.com/ plus an invoice reference, the assistant asks for the reference and adds it to the link.
A quick-reply button runs one of your bot replies when the customer taps it, so before it sends, the assistant asks which reply each button should run, for example Chat with robot or one of your own keyword-reply flows, and shows you the choices from your account. You can skip it: the message is still delivered, but tapping its buttons then starts no bot reply, and the assistant tells you so.
Yes. Tell it what the message should say and it drafts the template, shows it to you, and submits it to WhatsApp once you agree. It can make a normal marketing or utility message (with a header, footer and buttons), a one-time-passcode message, or a carousel of image or video cards. A new template cannot be sent until WhatsApp approves it, so the assistant then checks its status for you. It cannot edit or delete a template, so to change one it creates a corrected copy under a new name. Product carousels, which are built from your catalog, still have to be created in ChatSyncs.
Ask the assistant. It checks live with WhatsApp and tells you whether the template is approved, still in review, or rejected, and if it was rejected, why and what to change. A template can be approved by WhatsApp and still not be sendable from here if its variables are not mapped in ChatSyncs. The assistant says so when that is the case.
Yes, for looking up and organising them: where an ability asks for a number, you can give a contact’s chat id as the contact list shows it, such as BD.2035698430387918. Sending a message still needs a real phone number.
No. If your account has one WhatsApp number it’s used automatically. With several, the assistant lists them and asks which you mean.
The assistant shows you the same message the API returns, for example “Subscriber limit has been exceeded.” Nothing is sent, and nothing is charged.
The four destructive abilities are marked so your AI app asks more firmly before running them. To rule them out entirely, switch delete_subscriber, mark_conversation_status, remove_labels and remove_sequences off in your AI app. See Limiting what the assistant can do.