# Vitals 1.7: an Agents tab, and what is left of your AI plans Canonical page: https://vitalsmac.com/changelog/1.7 By Atta Ur Rehman Shah, maker of Vitals. Published October 14, 2026. **In short:** Vitals 1.7 adds an Agents tab. It lists every coding agent running on your Mac, Claude Code, Codex, Gemini CLI and the rest, with the project each works in, whether it is working or waiting for you, and everything it started. Above the sessions are your AI plans: for Claude, Codex, GitHub Copilot, Cursor, Antigravity, Grok, Ollama and OpenCode, how much of each limit is left, when it resets and whether you are on pace, for every account signed in on the Mac. Claude Code and Codex get a count of the tokens they used today and this week. And what is left of a plan can sit in the menu bar, in the dropdown and on the Overview, like any other reading. ## Your coding agents, as sessions Open the **Agents** tab and every coding agent running on your Mac is a row: the project it works in, whether it is working, waiting for you or idle, and for how long. Open a session and everything it started is listed under it, the dev server, the test run, the browser it drives, the tool servers it talks to, each with its CPU and memory. **Quit Session…** quits the agent and everything it started together, so nothing is left behind holding a port. Vitals asks first. An agent that is waiting for you is marked, in the window and in the dropdown, so you can see it without switching to the terminal. ## What is left of your AI plans Every AI plan has limits: a 5-hour session, the week, a month, sometimes a limit for one model. The Agents tab shows each as a ring with what is left in the middle, when it resets, and whether you are on pace or will run out before then. The week leads, since it is the limit that matters most. The 5-hour session and a model’s own limit follow as rows. 1. Open Settings → Integrations and turn on **Usage Limits**. 2. Sign in to the tools as you normally do. Vitals finds the sign-ins of Claude Code, Codex, GitHub Copilot, Cursor, Antigravity, Grok, Ollama and OpenCode on this Mac. A tool with no sign-in is left out. 3. Open the Agents tab. A limit that will run out before it resets says so in red. Vitals asks each maker for your usage the way the tool does itself, and keeps no copy of the sign-in. Nothing else is sent anywhere. ## Every account, every profile If more than one account is signed in on this Mac, each gets a ring of its own. Every Claude Code profile is found and named, and the Claude card shows them side by side, each with its own week, session and reset. In the dropdown, pick the account to look at from a list that says what each has left. For Claude Code inside the Claude app, turn on **Claude App Sign-In** in Settings → Integrations as well. ## Tokens For Claude Code and Codex, Vitals counts the tokens used today and this week from the logs the tools keep on your Mac: sent, received, read from the cache, and the total, with a chart of the week and the models that took them. Nothing is asked of anyone for this; it is all read here. ## Agents in the menu bar What is left of a plan can be a reading in the menu bar, like CPU or memory: each provider after its own mark, and the 5-hour session beside the week if you want it. Add it in Settings → Layout → Menu Bar. It turns red when a limit is nearly used up. ## The dropdown’s Agents tab The dropdown has an Agents tab too: one provider at a time, picked by its mark, with what is left of the week as the big number, the other limits as rows, the tokens of the week, and the sessions with a word on what each is doing. Click a row to open the window on it. ## Agents and Containers on the Overview The Overview in the window has an Agents card, and the dropdown’s Overview has two new tiles, Agents and Containers, in the grid and in the list alike. The Agents tile shows the plan nearest its end with a bar per provider, or the sessions if Usage Limits is off. Hide or move them like any other tile. ## Ask your assistant If you let Claude, Codex or Cursor ask Vitals, they can ask about your agents too: which sessions are running, what each started, and what is left of your plans. - “Which of my AI plans is closest to running out?” - “Is Claude Code waiting for me anywhere?” - “What did the Codex session in storefront start?” ## Questions people ask ### Which tools does Vitals see as agents? Claude Code, Codex, Gemini CLI, Aider, OpenCode, Cursor, GitHub Copilot, Amp, Kiro and Goose. A session is a running agent, with everything it started counted under it. ### Which plans can it read? Claude, Codex, GitHub Copilot, Cursor, Antigravity, Grok, Ollama and OpenCode, for every account signed in on this Mac. Turn on Usage Limits in Settings → Integrations. ### Does Vitals send anything anywhere? It asks each maker for your usage the way the tool itself does, with the sign-in the tool keeps on this Mac, and keeps no copy of it. Tokens are counted from the logs on your Mac. Nothing else is sent. ### Why does it say “Not signed in on this Mac”? Vitals found no sign-in for that tool here. Sign in with the tool once and it appears. For Claude Code inside the Claude app, turn on Claude App Sign-In as well. ### Why is the week the big number, not the 5-hour session? The week is the limit that matters most: a session comes back in hours, the week does not. The session and a model’s own limit are right under it, and either turns red when it is about to run out. ### Do the tokens include the Claude app? No. Tokens come from the logs Claude Code and Codex keep on your Mac, so they count what those two used. ### Can I hide the Agents tab? Yes. Tabs are arranged in Settings → Layout, and the tab only appears on a Mac that has an agent tool or a signed-in plan. ## Everything in 1.7 - New: An Agents tab. Every coding agent running on your Mac, Claude Code, Codex, Gemini CLI and the rest: the project it works in, whether it is working, waiting for you or idle, how long it has run, and everything it started, from the dev server to the tests. Quit a session and everything it started, together. - New: What is left of your AI plans. For Claude, Codex, GitHub Copilot, Cursor, Antigravity, Grok, Ollama and OpenCode: how much is left of the week, the 5-hour session and a model’s own limit, when each resets, and whether you are on pace or will run out first. For every account signed in on this Mac, each Claude Code profile included. Turn on Usage Limits in Settings → Integrations. - New: Tokens. How many tokens Claude Code and Codex used today and this week, where they went and which models took them, counted from the logs the tools keep on your Mac. - New: Agents in the menu bar and the dropdown. What is left of a plan as a reading in the menu bar, each provider after its own mark, with the 5-hour session beside it if you want it. The dropdown has an Agents tab with the plans, the tokens and the sessions. The window’s Overview gets an Agents card, and the dropdown’s Overview gets Agents and Containers tiles. - Improved: The strain notice can be hidden. The dropdown’s line about what is under strain stays on the Overview. Hide it with a right-click, or in Settings. - Fixed: Opening the GPU tab again shows the apps it last measured, not “Measuring…”, until the new reading is in. - Fixed: The temperature graph in the menu bar shows the last readings, not the current one repeated.