Let your assistant ask Vitals instead of guessing.
Claude, Claude Code, Codex, Cursor and any assistant that speaks MCP can ask Vitals what is running, which app is behind a number, what is on a port, and what the last month looked like. It gets real measurements instead of a process list, and it cannot quit anything without asking you.
Xcode is your build. Shall I ask Vitals to quit the dev server?
Done. 458 MB and port 4322 are free.
Zoom used the most, through your call at ten. Chrome is next, with 42 tabs open. Nothing here needs quitting.
A dev server in billing-api, started two days ago and idle for three hours. It still answers, so it is yours to keep or quit.
Left running.
The numbers are simulated: a web page cannot read your Mac, and should not.
One click per assistant, and it is on.
Claude, Claude Code, Codex and Cursor each have an Add button in Settings → Integrations that puts Vitals in the assistant's own list of tools. Any other assistant that speaks MCP takes the command shown there. Nothing is on until you add it.
- Add for Claude, Claude Code, Codex and Cursor
- A command for any other assistant that speaks MCP
- Switches for what every assistant may see
- Project folders by name only, if you prefer
It gets the same numbers you see, with when they were measured.
Left to itself, an assistant runs ps and guesses. Through Vitals it gets the picture you have: apps with their helpers added up, projects with their ports, sessions with what they started, and every answer stamped with when it was taken.
- What is using the CPU, memory, battery, disk or network
- Which app a number belongs to
- What is on a port, and whose it is
- Which coding sessions are running, and what the last month looked like
It asks Vitals, and Vitals asks you.
An assistant cannot quit anything on its own. The request comes to you in a window of its own: what is asking, what it wants to quit, what is connected to it right now, and whether Vitals thinks it looks safe. You click, or you don't.
- What is asking, and what it wants to quit
- What is connected to it right now
- Whether Vitals thinks it looks safe
- In folders you trust, safe requests can go through without asking
Honest about what the assistant hears.
Vitals itself sends nothing anywhere. What it tells an assistant goes wherever that assistant sends its conversations, and app names, project folders and ports are in there. You turn it on knowing that, and you choose what it may see.
- Vitals sends nothing anywhere
- App names, project folders and ports are in the answers
- Every tool but quit is read-only
- Off until you turn it on
MCP server, before and after
Ask an assistant why your Mac is slow and it runs ps aux, gets a thousand lines, and guesses. It sees node at 180% and cannot tell which project, which chat, or whether that is the test run you started on purpose. It sees a port in use and offers to kill whatever holds it.
Assistants are good at the next step, what to do about it. What they lack is the picture Vitals already has: apps with their helpers added up, projects with their ports, sessions with what they started, and what the last month looked like.
Vitals is an MCP server. Claude, Claude Code, Codex, Cursor and any assistant that speaks MCP can ask it what is running and get the same numbers you see, each with when it was measured. When an assistant wants something quit, it has to ask Vitals, and Vitals asks you, in a window of its own, with what is connected to it right now. You decide what it may see: system readings, apps, projects and history each have a switch, and project folders can go by name only.
Where it is in Vitals
The tab, the menu bar item and the setting, in the order you would reach for them.
- 01
Turn it on
Settings → Integrations lists Claude, Claude Code, Codex and Cursor with an Add button each. Add puts Vitals in the assistant's own list of tools; an assistant that is open sees it when it next starts. Any other assistant that speaks MCP takes the command shown there.
- 02
Choose what it may see
Switch off system readings, apps, projects or history for every assistant at once, and give project folders by name only, so an assistant hears "shop" and not where that folder lives on your Mac.
- 03
Ask
What is eating my memory. What is on port 8080. Which coding sessions are running, and which chat is each. How hard was the Mac pushed this week, and when was it at a limit. Every answer says when it was measured.
- 04
When it wants to quit something
The request comes to you in a window of its own: what is asking, what it wants to quit, whether that looks safe, and what is connected to it right now. Nothing closes until you say so. For a project folder you trust, add it in Settings → Integrations and safe requests there go through without asking.
MCP server: Activity Monitor and Vitals
The whole comparison is row by row. These are the rows about MCP server.
The words behind it
What the numbers mean, from the Activity Monitor glossary.
- Listening port
A listening port is a network port where a process is waiting for incoming connections, such as a web server on port 3000.
- PID
A PID, or process ID, is a number macOS assigns to each running process to identify it.
- Force Quit
Force Quit ends an app or process immediately, without letting it save or clean up.
- Helper processes
Helper processes are separate processes an app starts to do part of its work, such as rendering web pages, running extensions, or handling graphics.
Questions
The ones people ask about MCP server before they buy.
Which assistants work with it?
Claude, Claude Code, Codex, Cursor and any assistant that speaks MCP. Settings → Integrations adds the first four with one button each; any other takes the command shown there.
Can an assistant quit apps through Vitals?
Only through you. It asks Vitals, Vitals shows you what it wants to quit in a window of its own, and nothing closes until you say so. If you say no, it stays running.
Does it send my data to the assistant's maker?
Vitals answers the assistant on your Mac and sends nothing anywhere itself. What the assistant does with the answer is up to the assistant, the same as anything else you tell it.
Is it on by default?
No. It stays off until you turn it on in Settings → Integrations.
What can I stop an assistant from seeing?
System readings, apps, projects and history each have a switch in Settings → Integrations, for every assistant at once. Project folders can be given by name only.
Can safe requests go through without asking me?
Yes, for folders you choose. Add a project folder, or a folder that holds your projects, in Settings → Integrations, and what Vitals judges safe there is quit without asking. Everything else is still asked about.
How do I know what else will stop working?
Before anything is quit, the window names what is connected to it right now, such as Google Chrome or a process in another project, so you can see what else would break.
Can my assistant ask about last week?
Yes. It can ask how hard your Mac was pushed over a day, a week or a year, when it was at a limit, and which apps used the most over 90 days, 6 months or a year.
Read on
Guides that use this, and how Vitals compares with the apps people buy for it.
More of Vitals
All featuresMenu bar monitor
CPU, memory, GPU, network, temperature and more, each its own item, red while under strain, with the apps behind every number one click away.
History
CPU, memory, energy, disk and network from the last hour to the last year, with the apps that used the most, hour by hour.
Alerts
An app keeping the CPU busy, growing in memory, or hammering the disk or network gets named before the fans tell you.
Get Vitals for $15 while the launch price lasts
50 left at this price. The regular price is $39. One payment, lifetime updates, 14-day refund.
Get Vitals for $15