Connect your own computer to CellCog so agents can read, edit, and run commands in your real projects, with every action classified by risk and gated by your approval settings.
Cowork lets a CellCog agent work directly on your computer during a chat. Instead of working only in its cloud workspace, the agent can:
Key principle: you stay in control. The agent classifies every command as safe, moderate, or dangerous, and your one approval level decides which of those run without asking. Everything above it waits for your decision in chat.
cellcog.ai/cowork. You can also reach it from the Cowork chip in the chat editor (the Set up or Manage link) or from your profile page. Opened from a chat, the page shows Back to chat, which takes you straight back to that chat.The page's Setup steps list shows the same steps as the Cowork card in a chat, with the one you are on open and "Step N of 3" beside the title. After you click Download, the CellCog Desktop box reads "Waiting for the app" until the app connects. The page is not tied to one chat, so the last step stays open there: go back to your chat and turn Cowork on in it.
cellcog.ai/desktop/connect in your browser, where you are usually already signed in to CellCogThe code is valid for 10 minutes. If it expires, click Add account in CellCog Desktop again for a new one. You never type your password into CellCog Desktop itself.
Connect works only from a browser on the same network as the computer showing the code. A code someone else sends you (by email, chat or an invitation) can't connect their computer to your account, so never connect a code you did not start yourself. If you use a VPN or iCloud Private Relay, pause it for this step.
Signed in with an older version? Your existing sign-in keeps working. CellCog Desktop shows a one-time Reconnect with your browser prompt; follow it to move that account to the browser connection.
One CellCog Desktop can hold several CellCog accounts, for example a personal account and a company account. Click Add account again and connect the second account the same way.
Every connected computer has a key on your API Keys page (profile menu, then API Keys, or cellcog.ai/profile?tab=api-keys), named "CellCog Desktop · " followed by the computer's name. Click Remove on that key to disconnect the computer from your account at once. Connecting the same computer again later creates a new key and retires the old one.
Removing an account inside CellCog Desktop only makes the app forget it on that computer. To cut the connection from CellCog's side as well, remove its key on the API Keys page.
These keys do not count toward your API key limit, and they cannot be renamed or rotated.
~) if there is none.While Cowork is not set up yet, the Cowork chip's panel opens with "Step N of 3" and a short strip of the steps (App, Sign in, Turn on), above the same panel as before. If Cowork is on but something broke, it reads "Cowork is paused · fix step N" ("Cowork is paused · the app is updating" while the desktop app updates itself). The first time (on an account that never had the app), after you click Download the panel says "Waiting for the desktop app" until the app connects.
When an agent asks for Cowork in a chat, it shows a Cowork card under its message. The card lists the three setup steps, Get CellCog Desktop, Open it and sign in as the account it names, and Turn on Cowork for this chat, ticks the ones that are done and opens the one you are on. The last step has the same folder field and a Turn on Cowork button. Once you click Download, the card moves to step 2 and waits for the app to connect. When an agent sends the link more than once, earlier cards fold to one line with Continue in the latest message.
The chip and the card also show the connection state: whether the desktop app is connected, whether it needs an update, and whether more than one desktop app is signed in with your account. If Cowork is on but the desktop app is not connected, the card reads "Cowork is paused" and marks that step in yellow, and the chip says so too: open the app and sign in as the account they name. While the desktop app updates itself, the card also reads "Cowork is paused" and continues by itself once the app is back. On a phone, the card shows the same steps: steps 1 and 2 happen on your computer, and step 3 works from the phone.
Tip: the working directory gives the agent your project context and is an input to how it classifies risk (see below).
When the agent needs your machine, it sends one or more commands (terminal commands, file reads, file writes, file edits). Each command is classified by risk:
Different chats run on your computer in parallel: each chat gets its own lane, and commands inside one chat still run one at a time, in order. The desktop app shows one "Running on this computer" card per active chat. The Parallel chats setting in the desktop app's footer caps how many chats may run at the same time (default 8, from 1 to 16); chats beyond the cap wait their turn, and the app shows how many are waiting. Set it to 1 if you prefer one chat at a time. An agent whose chat is waiting behind others keeps waiting for its results instead of failing while the app is still going to run them.
The approval panel shows each pending command with:
You can Approve or Deny each command (optionally with a reason the agent will read), or use Approve All / Deny All for the whole batch. A batch starts executing once every command in it has been decided. In a regular chat the agent waits up to 24 hours for you. If 24 hours pass, or you send a new message instead of deciding, the waiting commands expire without running, and the agent checks again before asking a second time.
Everyday Cowork examples:
| Level | Cowork examples |
|---|---|
| Safe | Listing files, running tests or builds, reading and editing files inside a version-controlled working directory |
| Moderate | Reading files outside the working directory, reading environment variables or shell history, editing files in a project with no version control and no backup |
| Dangerous | Installing packages, deleting files, editing shell profiles or system files, pushing to remote repositories, anything touching production |
The full model, including how AI employees handle approvals, is in the Approvals & Safety Guide.
Cowork uses the same risk model and the SAME level as Browse my Chrome and connected Tools: the agent classifies, one level gates every surface.
Set your default in Profile > Approvals, or for one chat (or one AI employee) with Auto-approve in the + menu of the message box. What each level means on your computer:
| Level | What it auto-approves on your computer |
|---|---|
| None | Nothing. Every terminal command requires your approval. |
| Safe | Reads and edits inside your working directory when the file is git-tracked, plus side-effect-free terminal commands. |
| Moderate | Also reads of files outside your working directory, modifications to non-git-tracked files, and dev-mode database operations. |
| Dangerous | Also writes outside your working directory without git, package installs, deploys, and production-touching commands. The most permissive setting. |
Use default in the + menu's Auto-approve list returns a chat to your default. New chats always start at your default.
The working directory tells the agent where your project lives, and recoverability decides how edits are classified.
AGENTS.md. If there is none, it asks how you back the project up and offers to record your answer there. Without a recovery story, in-directory edits are classified moderate.~/.zshrc, /etc/hosts) are always dangerous to modify, and anything touching production is always dangerousYou can change the working directory mid-chat from the Cowork chip.
~/.ssh, ~/.gnupg, ~/.aws, ~/.kube, cloud CLI configs, ~/.npmrc, ~/.pypirc, ~/.netrc, .env files in your home directory, and your shell profiles.Cowork is the foundation; Browse my Chrome extends the same desktop app to your real Chrome browser. With a small Chrome extension, your agent can use the sites you are already logged into, gated by the same approval model. Instead of a working directory on disk, the trust boundary is a cyan "CellCog" tab group in your Chrome window.
See the Browse my Chrome Guide for setup. AI employees can also browse without your machine through their own Cloud Browsers.
AI employees can use Cowork when you enable it for their chat. An employee has one approval level for everything it does, like any chat: None, Safe, Moderate or Dangerous. A new employee starts at your own default level; change it at hire time or later. Actions above that level run only with a standing approval you have granted, or with your live approval while you are present; if you do not decide within 15 minutes the action fails and the employee's shift continues. Details are in the Approvals & Safety Guide.
If CellCog Desktop shows Connected but the chat or the Cowork page says it is not connected, the app is signed in to a different CellCog account than your browser. The chat and the Cowork page name the account your browser is signed in as. Open Accounts at the bottom of CellCog Desktop and switch to that same account. If it is not listed there, add it with Add account.
Commands run in your login shell, so tools configured in your shell profile (nvm, pyenv, conda, and similar) are normally found. If a tool is missing:
~/.zshrc or ~/.bash_profile)Each command runs as a separate process, so cd and environment changes do not carry over between commands.
/Users/me/projects/app, not ~/projects/app)No. You must install the desktop app, connect your account to it from your browser, and enable Cowork for a chat. Within a chat, only commands at or below your threshold run without a click.
The agent loses access immediately and pending commands fail. Reopen the app to resume.
Files the agent reads and command output it receives are sent to CellCog so the agent can work with them, and become part of that chat's context. Data is encrypted in transit; credential paths are blocked and secrets in output are redacted.
You can install the desktop app on several computers, but commands run on one computer at a time. With two open at once, choose one with Use this computer in CellCog Desktop (or in the chat); the other waits. Each computer keeps the same identity when you reinstall or update the app.
Yes. Click Add account in CellCog Desktop for each account. Each account's agents run under that account only. See "Several Accounts on One Computer" above.
The Cowork page offers builds for macOS, Windows, and Linux.
Only if the command is approved. Installing software is classified dangerous, so unless your threshold is Dangerous you review it first.
Remove the computer's "CellCog Desktop" key on your API Keys page: that disconnects it from your account at once, even if the app is still running. You can also quit CellCog Desktop and uninstall it.
No. Cowork connects your machine to your own CellCog account. Organization members cannot reach your machine through CellCog.