# Connect Ondes PRO to JetBrains AI Assistant

> Add the Ondes PRO MCP server to AI Assistant in IntelliJ IDEA, PyCharm or PhpStorm: the MCP setting, an API key, and what JetBrains does and doesn’t document.

Updated on 2026-10-21 · The Ondes PRO team · https://ondes.pro/en/integrations/mcp/clients/jetbrains

In JetBrains IDEs — IntelliJ IDEA, PyCharm, PhpStorm, WebStorm and the rest — MCP servers are used by **AI Assistant**. It accepts a remote server declared by its address. This page gives you the configuration for your Ondes PRO workspace and is upfront about which part JetBrains documents and which part it does not yet.

## Add the Ondes PRO MCP server to AI Assistant

You need a JetBrains IDE with the AI Assistant plugin enabled.

1. **Open AI Assistant’s MCP settings** — In **Settings**, go to **Tools → AI Assistant → Model Context Protocol (MCP)**.
2. **Add a server from its JSON configuration** — Click **Add** and choose JSON input. The form JetBrains documents for a remote server is just an address:

```json
{
  "mcpServers": {
    "ondes": { "url": "https://your-workspace.ondes.pro/mcp/" }
  }
}
```
3. **Pick your authentication** — If your AI Assistant version opens a window from your Ondes PRO workspace to authorise the connection, the form above is enough. Otherwise, add the `Authorization` header with an API key, as in the configuration below.
4. **Apply and check the status** — Apply the settings: the `ondes` server appears in the list with its status. Once connected, its eight tools are available in AI Assistant chat.
5. **Query your workspace from the IDE** — In AI Assistant chat, ask something like “Which actions does the invoices resource offer?”. The assistant calls the server’s tools to answer.

## API key configuration for AI Assistant

This is the path most likely to work with AI Assistant today. Create a key under **API & Developers → API keys** and give it only the scopes you need — invoicing read access, for instance.

Where to set it up : Settings → Tools → AI Assistant → Model Context Protocol (MCP).

**API key connection** (JSON configuration)

```json
{
  "mcpServers": {
    "ondes": {
      "url": "https://your-workspace.ondes.pro/mcp/",
      "headers": {
        "Authorization": "ApiKey opk_…"
      }
    }
  }
}
```

_Standard MCP compatible. Vendor documentation : https://www.jetbrains.com/help/ai-assistant/mcp.html_

> **What JetBrains documents, and what it doesn’t** — As of 21 October 2026, the AI Assistant documentation describes declaring a remote server by its `url`, but **neither sending `headers` nor OAuth authorisation** for such a server. The `headers` form above follows the standard MCP configuration used by other editors; it is offered as a fallback, with no guarantee. If your version ignores it, check JetBrains’ documentation, which moves fast. And if you go through a local bridge such as `mcp-remote`, use a version patched for CVE-2025-6514.

## Why a scoped API key suits an IDE

An OAuth connection carries your full identity in Ondes PRO: whatever your role allows, the assistant can do. In an IDE, where you often work on one slice of an integration, a **scoped key** is a sound habit: it only reaches the modules and actions you pick — `billing:read` to build an accounting export, say — and never goes beyond its holder’s permissions.

The key is shown only once, at creation. It can be suspended or revoked at any time from your workspace, and its calls are logged. Its per-minute quota is set when you create it. The [MCP server security](https://ondes.pro/en/integrations/mcp/security) page compares both options in detail.

## Example: coding customer returns without inventing an endpoint

_Example conversation · demo data (Atelier Lumen)_

> **You** : In `ReturnService`, when a return is accepted we need to cancel the matching invoice. Which Ondes PRO API call should I make?
>
> 🔧 `ondes_impossible` An issued invoice is neither deleted nor cancelled through the API: a credit note is issued.
>
> 🔧 `ondes_ressource` `billing/factures` sheet: `avoir/` action with a required `motif`, `nature` `total` or `partiel`, plus `montant_ht` and `taux_tva` for a partial one.
>
> **Assistant** : An issued invoice is not cancelled: you issue a **credit note**. For two lamps returned from a larger invoice, that is a partial one. I added a `POST billing/factures/{id}/avoir/` call to `ReturnService` with `nature: partiel`, the two lamps’ amount excluding VAT and the reason “Customer return”. The credit note gets an AV-2026-… number, and the total credited can never exceed the invoice amount. I did not issue anything from the IDE: invoicing has no test mode, so when to try it is your call.

## What AI Assistant gains from the Ondes PRO server

- Documentation for **your** workspace, current on every call, instead of a file copied into the project.
- A guard against hallucination: `ondes_impossible` lists what the API deliberately does not do.
- Complete resource sheets: typed and required fields, filters, ordering, actions, required permission.
- Only about 2,700 tokens of permanent context, versus roughly 44,000 for the old documentation file.

## Going further

- [Invoicing](https://ondes.pro/en/integrations/mcp/modules/invoicing) — Quotes, invoices, reminders and credit notes prepared by your assistant, within invoicing rules.
- [Build an integration](https://ondes.pro/en/integrations/mcp/guides/build-an-integration) — Scoped keys, signed webhooks, idempotency and test mode for a back office that never creates duplicates.
- [REST API v1](https://ondes.pro/en/integrations/rest-api) — Around 135 REST resources, scoped keys, an OpenAPI schema and signed webhooks.
- [MCP server security](https://ondes.pro/en/integrations/mcp/security) — OAuth 2.1, short-lived tokens, instant revocation and write safeguards: what your IT team needs to know.

## Frequently asked questions

### Where do I add an MCP server in a JetBrains IDE?

In Settings → Tools → AI Assistant → Model Context Protocol (MCP), with the Add button. For Ondes PRO, declare the address `https://<your-workspace>.ondes.pro/mcp/` in the `url` field.

### Does AI Assistant handle the OAuth connection with Ondes PRO?

JetBrains does not currently document OAuth authorisation for a remote MCP server. If your version opens an authorisation window, the connection should work; otherwise, use an API key in the `Authorization` header.

### Does AI Assistant support the headers field?

It is not described in JetBrains’ documentation as of 21 October 2026. We offer it because it belongs to the standard MCP configuration used by other editors, but we cannot guarantee it for AI Assistant.

### Which JetBrains IDEs does this cover?

Those that offer the AI Assistant plugin with MCP support: IntelliJ IDEA, PyCharm, PhpStorm and WebStorm, among others. The setting sits in the same place in each.

### Has the Ondes PRO team tested JetBrains AI Assistant?

No. AI Assistant implements the standard MCP protocol our server follows. Only Claude has been verified end to end by our team; for AI Assistant, authentication depends on what your version supports.