# AI use cases for Muxin (LumaryLabsLLC/muxin)

Repo: [LumaryLabsLLC/muxin](https://github.com/LumaryLabsLLC/muxin) | Use cases: 4 | Page: https://dedupedai.com/use-cases/lumarylabsllc-muxin/

Muxin is a free, open source desktop app and command-line tool that gathers the machines you own (Mac, Linux and Windows) into one hub for agent sessions, remote desktops, files, logs, GitHub work and self-hosted runners.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Check and fix self-hosted runners across your machines: https://dedupedai.com/use-cases/lumarylabsllc-muxin/check-and-fix-self-hosted-runners/
- Hand a GitHub issue to an agent on a chosen machine: https://dedupedai.com/use-cases/lumarylabsllc-muxin/hand-github-issue-to-agent-on-chosen-machine/
- Run one command on several machines at once: https://dedupedai.com/use-cases/lumarylabsllc-muxin/run-one-command-on-many-machines/
- Compare Claude Code and Codex usage and cost: https://dedupedai.com/use-cases/lumarylabsllc-muxin/check-claude-code-and-codex-usage-and-cost/

## How to check and fix self-hosted runners across your machines with Muxin

Repo: [LumaryLabsLLC/muxin](https://github.com/LumaryLabsLLC/muxin) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README

See which runners are busy, idle or offline, then install or restart them from one place.

**Who it is for:** A developer or DevOps engineer who runs GitHub self-hosted runners on several machines and gets jobs stuck in the queue.

**The problem:** Runners on different machines fail quietly, and it is hard to tell which one is offline or which queued job no runner can take.

**The outcome:** You see the state of every runner and the queued jobs no runner can take, and you can install, start, stop or remove runners on the machine you pick.

### Steps

1. Run `muxin --demo` to explore simulated machines first; nothing real is touched.
2. Run `muxin add-machine --name "Mac Studio" --host ci@mac-studio.local` to add a real machine.
3. Run `muxin watch acme` to see live runner status and jobs for the acme org.
4. Run `muxin install mac-studio --count 2` to install two runners registered with acme as services.
5. Run `muxin status` and check that each runner shows as working, idle or offline.

### Prompt to paste into your agent

```text
Using the Muxin repo at LumaryLabsLLC/muxin, add the machine <machine-host> with the Muxin CLI, then run `muxin status` and report which runners are working, idle or offline. Verify that the output matches what the Overview and Jobs pages show in the app.
```

### Caveat

Muxin reaches machines first with your system ssh, so each machine must accept your ssh key before any runner command can work.

Risk, credentials: Installing runners registers them with your GitHub organization, so use an account and org you control and keep ssh keys limited to machines you own.

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/lumarylabsllc-muxin/check-and-fix-self-hosted-runners/

## How to hand a GitHub issue to an agent on a chosen machine with Muxin

Repo: [LumaryLabsLLC/muxin](https://github.com/LumaryLabsLLC/muxin) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README

Pick an issue in Muxin and start an agent on the machine you choose to work on it.

**Who it is for:** A maintainer who triages issues and wants an agent to start work on a specific machine that has the right tools or hardware.

**The problem:** Starting an agent on the right machine for an issue means switching between GitHub, a terminal and the machine by hand.

**The outcome:** An agent session starts on the machine you picked, working on the issue you handed over, and you can follow it in the sidebar.

### Steps

1. Open Muxin and sync GitHub with ⇧⌘R, after your gh login is in place.
2. Open the GitHub page and find the issue or pull request in your inbox or repository list.
3. Click the button to hand the issue to an agent and choose the machine you want it to run on.
4. Watch the session in the sidebar, where a spinner shows while the agent works and a colored dot shows when it wants you.
5. Review the agent's changes in the workspace or pull request before you merge anything.

### Prompt to paste into your agent

```text
Using the Muxin repo at LumaryLabsLLC/muxin, hand the GitHub issue <issue-url> to an agent on the machine <machine-name> from the GitHub page in the app. Report the session name and the machine it started on, then confirm the session appears under that machine in the sidebar.
```

### Caveat

The README says GitHub access uses your gh login, so the issue hand-off works only after gh is signed in on the computer running Muxin.

Risk, credentials: The agent runs on your machine with your gh login and ssh access, so only hand issues from repositories you trust to that machine.

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/lumarylabsllc-muxin/hand-github-issue-to-agent-on-chosen-machine/

## How to run one command on several machines at once with Muxin

Repo: [LumaryLabsLLC/muxin](https://github.com/LumaryLabsLLC/muxin) | Difficulty: Beginner | Time: 30 minutes | Basis: From the README

Send one shell command to a group of machines and read each output side by side.

**Who it is for:** A person who looks after a few personal or office machines and needs the same check or update run on each.

**The problem:** Repeating the same command over ssh on every machine is slow, and the outputs get mixed up.

**The outcome:** Each machine's output and exit status appear side by side as the command runs, so you can spot the one that failed.

### Steps

1. Make sure each target machine is added in Muxin and reachable over ssh.
2. Open the Run page with ⌃6 (Alt on Linux and Windows).
3. Type the command, for example `df -h /` or `brew upgrade`, and select the machines to run it on.
4. Read each machine's output as it arrives, and check the exit status shown for each one.
5. To run it on just one machine, right-click that machine in the sidebar and choose Run a command.

### Prompt to paste into your agent

```text
Using the Muxin repo at LumaryLabsLLC/muxin, run the command `df -h /` on the machines <machine-list> from the Run page. Report each machine's disk output and exit status, and flag any machine with a failed exit status.
```

### Caveat

Run uses your login shell from your home folder on each machine, so commands must work there, and machines that are offline will show as unreachable instead of running.

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/lumarylabsllc-muxin/run-one-command-on-many-machines/

## How to compare Claude Code and Codex usage and cost with Muxin

Repo: [LumaryLabsLLC/muxin](https://github.com/LumaryLabsLLC/muxin) | Difficulty: Beginner | Time: 30 minutes | Basis: From the README

See tokens, requests and API-equivalent cost for both agents across all your machines.

**Who it is for:** A developer or team lead who uses Claude Code and Codex on several machines and wants to know where the usage goes.

**The problem:** Usage is spread across machines and logs, so it is hard to see which models, folders or machines drive the most work and what it would cost via the API.

**The outcome:** You see tokens and requests by day, model, machine and folder, plus what the same use would cost at Anthropic's and OpenAI's API list prices.

### Steps

1. Make sure Claude Code and Codex have run on the machines you added to Muxin, so their logs exist there.
2. Open the Usage page with ⌃5.
3. Choose the range: today, the last 7, 30 or 90 days.
4. Read tokens and requests by model, machine and folder.
5. Compare the API list price figure with your plan, remembering that with a Claude or ChatGPT plan you do not pay per token.

### Prompt to paste into your agent

```text
Using the Muxin repo at LumaryLabsLLC/muxin, open the Usage page for the last 30 days and report total tokens, requests and API-equivalent cost for Claude Code and Codex by machine. Verify the totals against one machine's logs before you give the final numbers.
```

### Caveat

Usage is read from the logs Claude Code and Codex keep on each machine, so machines not added to Muxin or with no logs will not appear in the totals.

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/lumarylabsllc-muxin/check-claude-code-and-codex-usage-and-cost/