OneLi with n8n
How to let an agent built with n8n place Hebrew phone calls through OneLi, follow the call, answer consultations and read the result.
n8n offers two paths: an MCP Client Tool node attached to an AI Agent node, which gets every OneLi tool, or an HTTP Request node that creates a call task directly.
Setup
- Create an API key at oneli.io/app/connect and store it on the server as
ONELI_API_KEY.
Checked against: MCP Client Tool node (node version 1.2+) and HTTP Request node
MCP Client Tool node (under an AI Agent)
Endpoint: https://mcp.oneli.io/mcp
Server Transport: HTTP Streamable
Authentication: Bearer Auth → credential token: oli_…
(or Header Auth: name "Authorization", value "Bearer oli_…")
Tools to Include: All
Options → Timeout: 60000 ms (default)HTTP Request node (or the AI Agent’s HTTP Request Tool)
Method: POST
URL: https://api.oneli.io/v1/call-tasks
Authentication: Generic Credential Type → Header Auth
Name: Authorization Value: Bearer oli_…
Send Body: on · JSON · Using JSON
Body: {"idempotency_key":"{{$execution.id}}","to_number":"+972…","goal":"…"}Notes
- Use a recent n8n: Streamable HTTP and the auth options need MCP Client Tool node version 1.2 or later.
- The official node page still describes version 1 (SSE Endpoint). The settings here come from the node source.
- An MCP OAuth2 option exists, but we have not verified whether it supports dynamic client registration. An API key is the tested path.
- After creating the task, a Wait node followed by GET /v1/call-tasks/:id returns the result. See the automation scenario on the use cases page.
Phone numbers: OneLi accepts an Israeli number in any common format (050-1234567, +972501234567). Once the call starts, the agent follows it with get_call_events or GET /v1/call-tasks/:id/events to answer consultations and get the summary.
Next steps
Sources: github.com/n8n-io/n8n/blob/master/packages/@n8n/nodes-langchain/nodes/mcp/McpClientTool/McpClientTool.node.ts · docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.httprequest/