Coding agent · JetBrains

Junie + SpendixAI

Junie CLI can use remote MCP servers over HTTP and shows an Authorization required status for servers that need sign-in. SpendixAI is added through the MCP screen or a JSON file.

Junie CLI installedA SpendixAI workspaceThe SpendixAI server address from Developer → MCP connections
What did I spend on food last month compared with the month before?
Junie · using SpendixAI · illustrativeIt looks at the records you allowed and answers from them. You stay in control of what is shared.
Read only to start · You approve what is shared

Connect in a few minutes.

Junie CLI adds SpendixAI from its /mcp screen or from a JSON file, then you authorize it in your browser.

Copy the server address

In SpendixAI open Developer → MCP connections and copy the SpendixAI server address.

Add the server

In Junie CLI run /mcp and press Ctrl+A to open the MCP Installation Assistant. If SpendixAI is not in the registry, press Ctrl+A again, choose a scope and the Remote connection type. Or save the JSON below in an MCP config file.

Authorize

The server shows Authorization required. In /mcp select it and choose Authorize. Finish the browser login, choose your workspace and pick a permission preset. Read only is the recommended start.

Check the status

Back in /mcp confirm the server shows Active, then ask a question about your spending.

mcp.json
{
  "mcpServers": {
    "spendixai": {
      "url": "https://spendix-api.kriva.studio/mcp"
    }
  }
}

The file is .junie/mcp/mcp.json for one project or ~/.junie/mcp/mcp.json for all projects. An optional timeout field is in milliseconds; the default request timeout is 60 seconds.

What you need

  • Junie CLI installed
  • A SpendixAI workspace
  • The SpendixAI server address from Developer → MCP connections
  • A browser for the Authorize step

Good to know

  • These steps are for Junie CLI. Whether Junie inside a JetBrains IDE supports OAuth sign-in for remote servers was not confirmed in the documentation we read.
  • The documentation we read did not describe whether Junie asks before it runs an MCP tool, so check that behaviour yourself before you rely on it. Keep the Read only preset.
  • A minimum Junie version or plan was not stated on the page we read.
  • A project-level file lives inside the repository, so prefer the user-level file for personal finance connections.

Questions to start with.

These need only read access. Add records or upload statements once you have moved to a broader preset.

“What did I spend on food last month compared with the month before?”“Which merchants did I pay most often this quarter?”“List my recurring charges and what each costs per year.”

Start read only. Widen it later.

Read onlyRecommended

Ask about transactions, summaries and reports. Nothing can be added, changed or uploaded.

Read and add

Ask questions, add expenses and upload statements or receipts.

Full access

Everything above, plus categorisation rules and insights. Billing and member settings stay in the console.

Junie and SpendixAI.

Does this work in the Junie plugin inside my JetBrains IDE?

We could not confirm it. The documentation we read describes OAuth sign-in for Junie CLI only.

Where do I see whether the connection worked?

Run /mcp in Junie CLI. A connected server shows Active. SpendixAI also lists connected assistants under Developer → MCP connections.

Do I need to paste an API key?

No. The connection uses your SpendixAI sign-in and a consent screen, with dynamic client registration, so there is no secret to copy into a config file.

Can I keep the connection read only?

Yes. Read only is the recommended preset, and you can grant more access later or disconnect the assistant at any time from Developer → MCP connections.

Setup steps were checked against JetBrains’ own documentation in November 2026. Menus and commands change, so if a step does not match what you see, follow the vendor’s current guide.

Using something else?

More assistants that can connect to SpendixAI. See all 42.

Bring Junie into your finances.

Create your SpendixAI workspace, import a statement, then connect the assistant.