# AI use cases for Lcu (amontlabs/lcu) Repo: [amontlabs/lcu](https://github.com/amontlabs/lcu) | Use cases: 4 | Page: https://dedupedai.com/use-cases/amontlabs-lcu/ LCU is a tool that lets agents in any harness use Codex's computer-use runtime to read windows, click, type, and take screenshots in desktop apps and Chrome, without Codex authentication. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Let Claude Code drive desktop apps with native approvals: https://dedupedai.com/use-cases/amontlabs-lcu/approve-desktop-apps-in-claude-code/ - Give Codex CLI control of desktop apps on macOS or Linux: https://dedupedai.com/use-cases/amontlabs-lcu/control-desktop-apps-from-codex-cli/ - Let an agent read and control Chrome tabs with site approval: https://dedupedai.com/use-cases/amontlabs-lcu/drive-chrome-tabs-with-site-approval/ - Run unattended computer use on a disposable VM or CI machine: https://dedupedai.com/use-cases/amontlabs-lcu/run-unattended-computer-use-on-a-vm/ ## How to let Claude Code drive desktop apps with native approvals with Lcu Repo: [amontlabs/lcu](https://github.com/amontlabs/lcu) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Install LCU as a Claude Code plugin so Claude can use desktop apps after you approve each one. **Who it is for:** A developer on an Apple Silicon Mac who uses Claude Code or the Claude app and wants the agent to operate apps like Zed, Notes, or Safari. **The problem:** The Claude app's Code tab used to fail app requests with 'Computer Use was not approved', and there was no clear way to grant access. **The outcome:** Claude Code can control approved desktop apps, and each app is approved by you through a native question card or terminal pane. ### Steps 1. Install the official ChatGPT desktop app and Claude Code 2.1.287 or later. 2. Run `claude plugin marketplace add amontlabs/lcu`. 3. Run `claude plugin install lcu@lcu`. 4. When the agent first controls an app, choose Allow this conversation, Always allow, or Deny on the approval card. 5. Run `/computer-use-apps` to review always-allowed apps and revoke any you no longer want. ### Prompt to paste into your agent ```text Using the amontlabs/lcu repo, install LCU as a Claude Code plugin on this Mac with the marketplace and plugin install commands from the README. Ask me to approve the first app the agent controls, then report the output of /computer-use-apps. Verify that the Approved apps panel lists only the apps I approved. ``` ### Caveat The Claude Code plugin install path is documented for Apple Silicon Macs and needs Claude Code 2.1.287 or later. Risk, credentials: Once an app is always allowed, the agent can act inside it, so avoid always-allowing password managers and browsers that hold logins. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/amontlabs-lcu/approve-desktop-apps-in-claude-code/ ## How to give Codex CLI control of desktop apps on macOS or Linux with Lcu Repo: [amontlabs/lcu](https://github.com/amontlabs/lcu) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README Register LCU with Codex CLI so the agent can see windows, click, type, and capture screenshots. **Who it is for:** A developer running Codex CLI on a Mac or Linux desktop who needs the agent to test or operate a GUI app that has no API. **The problem:** Codex CLI cannot interact with desktop windows by itself, so GUI-only steps need manual work. **The outcome:** Codex CLI can read windows and take screenshots in a desktop app, and each app asks for approval before the agent touches it. ### Steps 1. Update Codex CLI before setup, as the README requires for hook support. 2. Download the archive for your platform and its matching `.sha256` file from the latest release, then verify the checksum and extract it. 3. Run `~/.local/share/lcu/current/bin/lcu setup --agent codex` from the installed path, since the installer does not add `lcu` to your PATH. (inferred, check the README) 4. Follow the setup prompts for desktop-readiness and permission guidance. 5. Ask Codex CLI to list the windows of a named app and take a screenshot to confirm the connection. (inferred, check the README) ### Prompt to paste into your agent ```text Using the amontlabs/lcu repo, set up LCU for Codex CLI on this machine by following docs/INSTALLATION.md. Ask me to approve the desktop app named , then have the agent list its windows and take one screenshot. Report the setup result and confirm the screenshot was taken. ``` ### Caveat The README asks for a Codex CLI update before setup, and Linux support requires an active X11 desktop because native Wayland is unsupported. Risk, credentials: The agent can see and type into any app you approve, including apps that show saved credentials or private messages. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/amontlabs-lcu/control-desktop-apps-from-codex-cli/ ## How to let an agent read and control Chrome tabs with site approval with Lcu Repo: [amontlabs/lcu](https://github.com/amontlabs/lcu) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Enable the official ChatGPT Chrome extension so an agent can list and control tabs, one approved site at a time. **Who it is for:** A researcher or operator who wants an agent to gather information from or act within sites already open in their Chrome profile. **The problem:** Agents cannot see the tabs you already have open, so routine web tasks need copy-paste between the agent and the browser. **The outcome:** The agent lists your Chrome tabs and works on sites you have approved, and it reports what it found. ### Steps 1. Install the official ChatGPT desktop app and enable the official ChatGPT extension in your Chrome profile. 2. Restart your harness, then restart Chrome or turn the ChatGPT extension off and on in `chrome://extensions`. 3. Ask your agent to use LCU to list Chrome tabs. 4. Approve each site when the agent asks for it. 5. If calls fail with `requires caller identity`, follow the Chrome troubleshooting section of the installation guide. ### Prompt to paste into your agent ```text Using the amontlabs/lcu repo, enable Chrome control in LCU following the Chrome section of docs/INSTALLATION.md. Ask me to approve one site, then have the agent list my Chrome tabs and summarize the title of the approved site. Verify that no site was accessed before I approved it. ``` ### Caveat Claude Code's Chrome support is experimental because some interruptions do not trigger tab cleanup. Risk, credentials: The agent works inside your logged-in Chrome profile, so approve only sites where you are comfortable with the agent acting for you. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/amontlabs-lcu/drive-chrome-tabs-with-site-approval/ ## How to run unattended computer use on a disposable VM or CI machine with Lcu Repo: [amontlabs/lcu](https://github.com/amontlabs/lcu) | Difficulty: Advanced | Time: 45 minutes | Basis: From the README Set LCU approval to auto on a sandbox machine so the agent can run without a person at the keyboard. **Who it is for:** A DevOps engineer or tester who runs agent-driven GUI checks on a VM or CI runner where no one can answer approval prompts. **The problem:** Interactive approval prompts block unattended jobs, so GUI test runs stall when no one is present. **The outcome:** The agent runs GUI steps on the disposable machine without prompts, and per-app approval still applies to every app it touches. ### Steps 1. Install the official ChatGPT desktop app and your harness on the disposable VM or runner. 2. Run `~/.local/share/lcu/current/bin/lcu setup --approval auto`. (inferred, check the README) 3. Run `lcu apps` to confirm which apps are always allowed on this machine. 4. Run the agent job, then review the results and any screenshots it saved. 5. Run `lcu cross-turn off` if you turned on cross-turn computer use for this job. ### Prompt to paste into your agent ```text Using the amontlabs/lcu repo, configure LCU for unattended use on this disposable machine with `lcu setup --approval auto`. Run `lcu apps` and report the list of always-allowed apps. Verify that the list contains only apps the job needs. ``` ### Caveat The auto approval mode only pre-allows LCU's two model tools in the harness, and per-app approval stays in force. Risk, remote script: Setup runs a downloaded installer and unattended approval lets the agent act without a person present, so use it only on disposable machines. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/amontlabs-lcu/run-unattended-computer-use-on-a-vm/