How to Connect Xentral to Gemini via MCP

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Xentral ERP, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

Xentral + Gemini

Connect Xentral ERP to your AI: search products and stock per warehouse and storage location, customers and suppliers, sales orders, delivery notes with tracking, invoices with payment status, credit notes, purchase orders and supplier invoices, and create sales order drafts. Current Xentral API with a Personal Access Token.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

XENTRAL_URL, XENTRAL_TOKEN

Step 1 — Get credentials

1. In Xentral, click Administration at the bottom left, then Account settings (Kontoeinstellungen) → Developer settings (Developer Einstellungen) → Personal Access Tokens. A quicker way: type "Personal Access Token" into the Smart Search. You need an admin account for this.
2. Click Create token (Token erzeugen) and give the token a name such as AnythingMCP-Read, a description and a lifetime (the default is 180 days).
3. Grant Read on the areas the AI should see: products and stock, customers (CRM), sales documents (sales orders, delivery notes, invoices, credit notes), purchasing (purchase orders, suppliers, liabilities) and settings such as payment methods and projects. Add Create and Delete on sales orders only if the AI may prepare order drafts. Leave out the Admin scopes.
4. Click Create and copy the token right away: Xentral shows it only once.
5. Set XENTRAL_URL to the address you open Xentral at, for example https://acme.xentral.biz (or your own domain), with no /api, no path and no trailing slash. Paste the token into XENTRAL_TOKEN on its own, without the word Bearer, then install the connector.
6. Start with xentral_list_products: if products come back, the URL and the token work.

Token lifetime and permissions: a token expires on the date you chose, and Xentral deletes tokens that have not been used for 180 days. Its permissions cannot be edited later: to change them, create a new token and replace XENTRAL_TOKEN. The AI sees only the areas the token may read.

Ids, not printed numbers: every tool takes Xentral's internal id, not the product, document or customer number printed on paper. The AI first finds the id with a list tool, through a filter on number, documentNumber or customerNumber, or through search. A customer's id is also its address id: documents show it as address.id, and xentral_create_sales_order takes it as addressId. Customer filters only match exactly, so to find a customer by part of the name the AI searches their documents with xentral_list_invoices or xentral_list_sales_orders and takes address.id from there.

Filters: list tools take conditions that must all match, each with a field (key), an operator (op) and a value. Open invoices, for example: [{"key": "paymentStatus", "op": "in", "value": ["pending", "partiallyPaid"]}]. Operators include equals, in, contains, startsWith, greaterThanOrEquals and lessThan; each tool lists the fields it accepts. Document dates are YYYY-MM-DD.

Searching and sorting documents: for sales orders, delivery notes, invoices, credit notes and purchase orders, search looks through the document number, customer or supplier name and number, email, ZIP code and external order numbers. sort takes field names, a leading minus sorts descending (-documentDate), and include embeds related data such as lineItems or address.

Statuses: sales orders are draft, released, completed or cancelled; invoices, delivery notes and purchase orders can also be sent. The payment status (paymentStatus) of invoices and credit notes is pending, paid, partiallyPaid, cancelled, partiallyCancelled, badDebt or debited.

Stock: xentral_get_product_stock returns totals and the same figures per warehouse and per sales channel. sellable is what can still be promised to a customer, physical what is on the shelves, reserved what open orders hold. Warehouses (Lager) contain storage locations (Lagerplätze): xentral_list_product_storage_locations shows where a product lies, xentral_list_storage_location_items what lies in one storage location.

Shipments and payments: xentral_list_delivery_note_shipments returns the carrier, tracking number and tracking link of a delivery note. xentral_get_invoice_balance shows the payments booked against an invoice and how much is still open.

Pages: document lists take page and perPage (up to 1000, 25 by default); the other lists take page and pageSize (10 to 50).

Writing: xentral_create_sales_order only creates a draft for an existing customer: no order number, and nothing is reserved or shipped until someone checks and releases it in Xentral. Prices the AI leaves out come from the customer's and the product's prices in Xentral. xentral_delete_sales_order_draft deletes such a draft and is switched off at install; Xentral refuses to delete released, completed or cancelled orders. Every other tool only reads.

Read-only setup: give the token only Read permissions and switch off xentral_create_sales_order on the connector page.

Rate limit: Xentral allows about 100 requests per minute per instance, whatever the plan; above that it answers 429 with a Retry-After header. One filtered list call is better than many single lookups.

Common errors: 401 means the token is wrong, expired or deleted. 403 means the token lacks the permission for that area: create a new token that has it. 400 "Generic request validation failed" or "filter must be an array" points to a filter field or operator the tool does not accept, or to a malformed value. 404 "Route not found" means XENTRAL_URL contains a path such as /api/v1, or the instance does not offer that endpoint yet. 409 when deleting a sales order means it is no longer a draft.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Gemini

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Gemini CLI reads MCP servers from ~/.gemini/settings.json (on Windows %USERPROFILE%\.gemini\settings.json). Add:

{
  "mcpServers": {
    "anythingmcp": {
      "httpUrl": "https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID",
      "headers": { "X-API-Key": "YOUR_MCP_API_KEY" }
    }
  }
}

1. Create an MCP API key in AnythingMCP under MCP Servers → your server → API keys, and paste it in place of YOUR_MCP_API_KEY.
2. Save the file and restart gemini.
3. Run /mcp inside Gemini CLI (or gemini mcp list in your shell): anythingmcp should show as connected.
4. Start prompting: Gemini CLI can now call the Xentral tools.

Available tools

Example prompts

• "How many units of product 700123 can we still sell in each warehouse, and on which storage locations does it lie?"
• "Which invoices are still unpaid or only partially paid? Show customer, amount and dunning level, oldest first."
• "Where is the parcel for order 200345? Give me the carrier and the tracking link."
• "Create a sales order draft for customer 10001 with 20 units of product 700123 and 5 units of product 700456, desired delivery date 20 October."

FAQ

Where does Gemini CLI look for MCP servers? In ~/.gemini/settings.json for your user, or in .gemini/settings.json inside a project. gemini mcp list shows each server and whether it connects.

Next steps

• Connector landing page
• Connect to Claude
• Connect to ChatGPT