AI agents

gluvyn has an MCP (Model Context Protocol) server. With it, agents such as Claude, Codex, Cursor and GitHub Copilot in VS Code read your month and record bills, payments, incomes and subscriptions in your account from a plain-language request.

How it works

The agent signs in with the same account as the app, Google or Apple, on the gluvyn sign-in page. You see the access request and approve it.
Every change the agent makes goes to your gluvyn cloud copy. The app gets it on its next sync: when you open the app, or every 30 seconds while it is open.
The app's rules apply: amounts, categories, installments and payments go through the same checks, and nothing is deleted.

Before connecting, sign in to the app with your account and wait for the first sync to finish. Without that cloud copy the agent records nothing, so it never conflicts with the data on your phone.

If you edit on the phone while offline as the agent records something, the app shows a conflict and you choose which copy to keep, just like with two devices.

Set it up

Use this address in your agent. The first time, it opens the browser so you can sign in and approve:

https://mcp.gluvyn.app/mcp

Claude Code

Run the command below in the terminal. Then, in Claude Code, type /mcp, choose gluvyn and sign in.

claude mcp add --transport http gluvyn https://mcp.gluvyn.app/mcp

Claude (claude.ai and desktop app)

In Settings › Connectors, choose Add custom connector, name it gluvyn and paste the address. Then choose Connect.

https://mcp.gluvyn.app/mcp

Codex

Run the commands below in the terminal. The second one opens the browser to sign in.

codex mcp add gluvyn --url https://mcp.gluvyn.app/mcp
codex mcp login gluvyn

Cursor

Add the server to ~/.cursor/mcp.json. In Settings › MCP, choose Login next to gluvyn.

{
  "mcpServers": {
    "gluvyn": {
      "url": "https://mcp.gluvyn.app/mcp"
    }
  }
}

VS Code (GitHub Copilot)

Create .vscode/mcp.json in the project, or use the MCP: Add Server command. VS Code asks you to sign in when the server starts.

{
  "servers": {
    "gluvyn": {
      "type": "http",
      "url": "https://mcp.gluvyn.app/mcp"
    }
  }
}

Other agents work if they support remote (HTTP) MCP servers with OAuth sign-in. Use the same address.

What the agent can do

get_month reads the salary, goals, incomes, bills and totals of a month.
list_subscriptions lists your subscriptions and the monthly and yearly totals.
add_expense adds a bill to a month, with category, minimum payment, installments and what was already paid.
pay_expense records a full or partial payment of a bill, in cash, Pix or by card.
add_income adds an extra income to a month.
add_subscription adds a monthly or yearly subscription, in reais or dollars.
update_month_plan sets the net salary and the savings and investment goals of a month.

Example requests

“Add the Nubank card bill of R$ 1,240.50 to November, minimum R$ 186.”
“I paid the power bill by Pix, mark it as paid.”
“I got R$ 800 from a freelance job today.”
“Add ChatGPT Plus, US$ 20 a month, with the dollar at R$ 5.40.”
“How much is left to pay this month?”

Privacy and security

The agent only reaches the account you approved and cannot delete data. Review what it plans to record before confirming: AI agents make mistakes.

To disconnect, remove the gluvyn server from your agent. To revoke every access you approved, write to support@gluvyn.app.

What the agent reads or records also goes through its provider. Details in the privacy policy.