Brunvo, inside your other AI apps.

Brunvo has its own MCP server. Connect Claude Code, Codex, VS Code, Claude Desktop, OpenCode, ChatGPT, or another AI app, and that app can see and run your desks, sessions, agents, issues, and Work Hub, start an agent, or send it a prompt from your library.

Connect an app, then ask it.

Pick the AI app in Settings ▸ Connected Apps. It asks Brunvo once, you approve it, and you choose what it may do. Then ask it to start an agent, check on one, or send it a prompt from your library.

Settings ▸ Connected Apps · brunvo mcp doctor checks the setup

Each app gets what you allow.

Access is chosen per app, and an app never raises its own. Brunvo keeps a record of what each one did, and you can revoke it at any time.

Can look
Sees your desks, sessions, agents, and issues. Changes nothing.
Can work
Also starts agents, sends them work, and stops them.
Full control
Everything you can do in Brunvo, including screens, closing, settings, and commands. It never asks first.

From the web, too.

ChatGPT and claude.ai reach your Mac through a private link while Brunvo is open, and reconnect on their own after a restart, a wake from sleep, or a change of network. What an app reads is cleaned of hidden characters and likely passwords first.

Access
Can look, Can work, or Full control, chosen per app; an app never raises its own
Record
See what each app did, and revoke it
On the web
ChatGPT and claude.ai reach your Mac through a private link while Brunvo is open
Clean text
What an app reads is cleaned of hidden characters and likely passwords

Every example here is a working simulation of Brunvo 0.2.7 on an example project: it plays by itself, and a click makes it yours. The agents are scripted; no AI model is involved.

More of Brunvo.

Try it for 7 days.

No card. The trial starts the first time you open Brunvo on your Mac and sign in.