In short
LlamaIndex agents use OmniChannels.ai through our remote MCP server. With llama-index-tools-mcp, a BasicMCPClient pointed at https://mcp.broadnet.me/mcp and wrapped in McpToolSpec turns our send, template and delivery report tools into agent tools. Our server requires an Authorization: Bearer header, so check the package's documentation for how your version sets it, or call our REST API from your own tools.
What you can do
Delivery support agent with read tools only
Keep only the read tools so the agent answers delivery questions without being able to send.
Try asking
“Build an agent that answers delivery questions using only read tools, and refuses any request to send a message.”
Answer from your docs, act through tools
Combine a query engine over your own policies with our messaging tools, so replies follow your rules before anything is sent.
Try asking
“Build an agent that checks our refund policy index, drafts a WhatsApp reply in Arabic and English, and asks me before sending it to 971500000000.”
Template-aware workflow
The agent lists approved WhatsApp templates and fills variables for business-initiated messages outside the 24-hour window.
Try asking
“Write a workflow that finds the approved appointment_reminder template and sends it to 966500000000 with tomorrow's time slot, after my approval.”
Narrow REST tools
Wrap a single REST operation in a FunctionTool when you want full control over fields, validation and logging.
Try asking
“Write a LlamaIndex FunctionTool that sends one SMS with POST /sms/sendSms and reads BROADNET_TOKEN from the environment.”
Failure analysis
The agent reads delivery reports and groups failures by reason for your operations team.
Try asking
“Create an agent that reads yesterday's delivery reports and summarises FAILED and REJECTED messages by reason and channel.”
How it connects
llama-index-tools-mcp provides BasicMCPClient, which connects to a remote MCP server by URL, and McpToolSpec, which turns that server's tools into a tool list for a LlamaIndex agent. Point the client at https://mcp.broadnet.me/mcp, a Streamable HTTP endpoint. Every request to our server must carry an Authorization: Bearer header; how you set it, for example through a custom HTTP client, depends on the package version, so check its documentation.
Filter the tool list so each agent has only the tools it needs, and add a confirmation step before any send. If the header option in your version does not fit, write your own FunctionTool wrappers around our REST API with a Bearer token, using the OpenAPI 3.1 spec for exact fields; single SMS is JSON, bulk and most other channel sends use multipart/form-data.
- MCP: BasicMCPClient at https://mcp.broadnet.me/mcp, wrapped in McpToolSpec
- Authorization: Bearer header required on every request
- Filter tools by name and confirm before any send
- REST alternative: FunctionTool wrappers built from the OpenAPI 3.1 spec
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 a temporary key
In the admin portal open Integrations, then MCP Integration, choose Get temporary key and set it as BROADNET_TOKEN.
Guide: MCP server for AI agents - 3
Install llama-index-tools-mcp
Install the package and check its documentation for how your version passes custom headers to a remote server.
- 4
Load the tools
Create a BasicMCPClient for our URL with the Authorization header, wrap it in McpToolSpec and get the tool list.
Guide: MCP server for AI agents - 5
Restrict and gate
Pass only the tools the agent needs to it, and require confirmation before any tool that sends.
- 6
Test on your own number
Run the agent against your verified number and check the delivery report before you widen access.
Vendor documentation
LlamaIndex is a trademark of its owner. OmniChannels.ai is not affiliated with or endorsed by LlamaIndex; this page describes a connection built with LlamaIndex's own MCP, HTTP or webhook features.
