n8n and Make
There is no private node package to install for the first workflow. Use the HTTP and MCP endpoints.
n8n
For an agent with tools, set the MCP client URL to https://mcp.maskio.io/v1/mcp/gateway/{id} and the bearer key to mp_live_....
For a single completion, add an HTTP Request node:
- Method
POST - URL
https://api.maskio.io/v1/proxy/chat/completions - Headers
Authorization,X-Muteproxy-Workspace,X-Muteproxy-Session - Body: the chat
messages
Prefer to persist the tokenized output. n8n’s own execution log is still n8n’s. Self-hosting n8n in the EU does not, by itself, stop a model node from receiving raw fields.
Make
Use the same HTTP call in an HTTP module before any native OpenAI module. If a later module must update the CRM, prefer an MCP write through the gateway so Make does not reassemble the raw record for the model.
Quotas are the text-page and MCP-call allowances on your plan.