In short
Cursor connects to OmniChannels.ai through our remote MCP server at https://mcp.broadnet.me/mcp. Add it to .cursor/mcp.json with your token in an Authorization header read from an environment variable. Cursor's agent can then send WhatsApp, SMS and RCS messages, list approved templates and read delivery reports, asking for approval before each tool call. No extension is needed.
What you can do
Send test messages while coding
Check a message flow end to end without leaving the editor, sending only to your verified numbers.
Try asking
“Send a WhatsApp message to 971500000000 saying the order has shipped.”
Look up template names and variables
The agent lists approved WhatsApp templates so the code you write uses the exact names and parameters.
Try asking
“List my approved WhatsApp templates in Arabic and show the variables each one expects.”
Debug delivery problems
Read the delivery report for a message your code just sent and see why it failed.
Try asking
“Check the delivery report for the last message I sent and explain any failure.”
Write integration code from the spec
Give the agent our OpenAPI spec so the REST calls it writes use the exact fields, including multipart sends.
Try asking
“Using the OpenAPI spec, write a TypeScript function that sends a bulk SMS campaign.”
How it connects
OmniChannels.ai runs a remote MCP server over Streamable HTTP. Cursor reads MCP servers from .cursor/mcp.json in a project or ~/.cursor/mcp.json for every project, and resolves ${env:NAME} in the url and headers fields, so the token can stay in an environment variable instead of the file.
Once connected, Cursor lists the server's tools in its MCP settings, where you can switch each one on or off. Cursor asks for approval before running an MCP tool by default and shows the arguments first. Keep approval on for every tool that sends, since each send is a real, billed message.
- Endpoint: https://mcp.broadnet.me/mcp (Streamable HTTP)
- Header: Authorization: Bearer ${env:BROADNET_TOKEN}
- Config: .cursor/mcp.json (project) or ~/.cursor/mcp.json (global)
- Troubleshoot in the Output panel under MCP Logs
Set it up
- 1
Request API access
Request test access and receive an API token. Test sends go only to your own verified numbers.
- 2
Get an MCP key
In the admin portal open Integrations, then MCP Integration, and choose Get temporary key.
Guide: MCP server for AI agents - 3
Add the server to mcp.json
Add an omnichannels entry under mcpServers with our URL and an Authorization header that reads ${env:BROADNET_TOKEN}.
Guide: Use the MCP server with Cursor - 4
Set the variable and restart
Set BROADNET_TOKEN in the environment Cursor starts from, such as your shell profile, then restart Cursor.
- 5
Review the tools
In Cursor's MCP settings, switch off every tool the agent does not need and keep approval on for send tools.
- 6
Send a first message
Ask the agent to send a test message to your verified number, approve the call, then check the delivery report.
Vendor documentation
Cursor is a trademark of its owner. OmniChannels.ai is not affiliated with or endorsed by Cursor; this page describes a connection built with Cursor's own MCP, HTTP or webhook features.
