MCP Server
The Nuntiq configuration server is an MCP server. It lets an implementation consultant configure a Nuntiq customer from an AI app such as Claude (Claude Code, Claude Desktop, claude.ai), Cursor or VS Code, instead of clicking through the admin portal.
The AI app works with your own admin portal account and rights, on the customers you are assigned to. It can:
- read a customer's configuration and explain it, find what refers to what, and compare two customers;
- test validation and enrichment rules, expressions, capture rules, workflow steps and inbound file formats against real invoices, render email templates and preview PDF templates;
- propose changes. Configuration changes come as a change set with a field-by-field diff. Nothing is changed until you approve it;
- manage users, groups, API users, webhooks, supplier integration access, CDN files, maintenance windows and Excel imports, each described first and carried out only after you confirm;
- run a scheduler task now and read what a connector run printed, to check and correct a connector set-up.
Every change shows in the customer's Change History as your name followed by
the app, for example Jane Smith (via Claude Code).
Environments
There is one server per environment. Each signs you in against that environment's admin portal, and a connection to one environment cannot read or change another.
| Environment | Server URL |
|---|---|
| Test | https://test-api.apreceiving.com/mcp |
| Production | https://api.apreceiving.com/mcp |
How it fits together
- The AI app talks to the MCP server over Streamable HTTP with an OAuth access token.
- You sign in once in the browser, on the normal admin portal sign-in page (password or single sign-on), and approve what the app may do.
- The MCP server never touches a database. Every call goes through the same admin API routes and rights checks as the admin portal.
What's in this section
- Before you start: your rights and the customer's consent.
- Connect your AI app: set-up per app.
- Your first session: what to ask first.
- Change sets: how changes are proposed, tested, approved and reverted.
- Guided prompts: ready-made work plans.
- Connectors through the MCP server: building and testing a connector from your AI app.
- Security and data: what the AI app sees and what stays in the portal.
- Tools, Resources, Troubleshooting.