# Agents That Use a Computer: approval gates, guard rules and status signals for desktop, browser and device agents

> Runbooks for agents that drive real machines: sort every action from read to irreversible, write a one-screen runbook per surface, add a guard that fails closed, answer approval cards in seconds and see when an agent is waiting on you.

- URL: https://paperchasewebb.ai/bots/marketplace/items/agents-that-use-a-computer-approval-gates-and-status-signals/
- Type: Prompt pack
- Category: Solo business
- By: PaperChaseWebb
- Works with: Claude Code, Codex, Grok Bot, Cursor (named in the listing); Plain .md
- Price: $29 USD
- Difficulty: Intermediate
- Setup time: 45 min
- Time to result: One surface with a runbook, a guard and a status signal in an afternoon
- Format: Markdown (.md) file, agents-that-use-a-computer-approval-gates-and-status-signals.md
- Version: 1.0, updated 2026-10-27
- License: Personal and commercial use on your own projects and your team's. Do not resell or republish this pack.

## What you end up with

Your agent works alone on everything it can undo, asks with evidence before anything it cannot, and never sits unnoticed on a permission prompt.

## What you get

- 15 sections in one .md file, about 14 pages
- A five-class action sheet from read to irreversible
- A surface ladder: API, generated CLI, browser, then pixels
- A one-screen RUNBOOK.md template
- A fail-closed guard layer with a bypass test list
- An approval card format and single inbox rules
- Runbooks for 3D printers, TVs, paid APIs, price watches and app releases
- Status signal states and the metadata-only rule
- A pre-flight checklist, a rules block, two SKILL.md files and 12 prompts
- Linked credits for the 12 public sources we studied

## Who it is for

People pointing agents at desktop apps, browsers, printers, TVs, payment endpoints or release pipelines who want the agent fast on safe work and stopped cold on anything it cannot undo.

## Tools needed

- Claude Code, Codex, Cursor or Grok: operates the surface and raises approval cards
- Agent hooks: guard checks and status signals
- Playwright or a device API: the surface the agent drives
- You need: An agent with shell or browser access
- You need: One surface you want it to operate (app, site, device, paid API or release pipeline)

Cost to run: Free. Uses the agent plan you already pay for. Optional hardware for status lights is listed so you can decide.

## Sample prompt: Sort the actions (Claude Code, Codex, Cursor, Grok)

```
List every action you can take on [surface] with the access you have. Sort each into read, draft, local reversible, external reversible or irreversible.
When unsure, mark irreversible and say why.
```

The agent may never argue an action down a class.

## Sample prompt: Attack the guard (Claude Code, Codex)

```
Write 20 test commands that try to get a dangerous action past this guard hook: trailing comments, script files, long padding, aliases, harmless-then-harmful chains.
Run them and report which got through.
```

Every bypass you find becomes a permanent test.

## Free preview of agents-that-use-a-computer-approval-gates-and-status-signals.md

````markdown
# Agents That Use a Computer: approval gates, guard rules and status signals for desktop, browser and device agents

A PaperChaseWebb pack, version 1.0. Written for Claude Code, Codex, Cursor and Grok, and plain Markdown, so the action sheet, the runbook template, the guard rules and the prompts work with any agent that can run commands, drive a browser or call a device API.

Agents no longer stop at writing code. People now point them at desktop apps through command line wrappers, at browsers, at 3D printers and TVs on the home network, at payment endpoints and at app store release pipelines. The jump in risk is not gradual. An agent that drafts a file can be wrong for free. An agent that starts a print, pays an invoice, submits a build or deletes a folder cannot take it back.

This pack gives you one split (reversible or not), one runbook format per surface, a guard layer that fails closed, approval cards you can answer in seconds and status signals so a waiting agent never sits unnoticed. It is built from public build reports by people running agents on real machines and devices, plus our own runbooks. Sources are credited by link in the Sources section.

## What is inside

| Section | What it is for |
|---|---|
| 1. The one split | Sort every action into five classes, from read to irreversible |
| 2. Pick the surface | Command line, API, browser or pixels, and why the order matters |
| 3. The runbook template | One RUNBOOK.md per surface: allowed, confirm first, never |
| 4. The guard layer | Pre-action hooks, allowlists, fail-closed checks and how to test them |
| 5. Approval cards | One format, one inbox, evidence attached before handoff |
| 6. Device runbooks | 3D printers, TVs and other things on your network |
| 7. Money runbooks | Paid APIs, spend caps and price watches that never rebook on their own |
| 8. Release runbooks | Remote fixes that stop at the test build |
| 9. Status signals | Lights, small screens and phone alerts for "needs you" |
| 10. One front door for many sessions | Routing tasks to live agent sessions without losing control |
| 11. Before the first unattended run | A checklist |
| 12. Rules block | A paragraph for CLAUDE.md, AGENTS.md or a Cursor rule |
| 13. Two skills | Runbook writer and approval card |
| 14. Prompts | 12 prompts |
| 15. Sources | What we used and how we used it |
````

## Worth knowing

- Hook names, permission modes and computer-use features change often. Check current docs for your agent.
- Guards reduce risk, they do not remove it. Keep backups, narrow keys and spend caps outside the agent.
- Not financial advice. Payment agents need rules from whoever owns the money.

## Questions

**Is this only for coding agents?** No. The action sheet, runbook and approval card work for any agent that acts on a computer, including browser agents and assistants with tool access. The hook examples assume a coding agent.

**Do I need the status light hardware?** No. A phone push or a chat message works. The pack lists light and small screen options for people who want one.

**Will it let my agent buy things?** It can fill carts and check prices. Paying, booking and checkout stay irreversible actions that need your approval every time.

**Is there a free sample?** Yes. The five-class action sorting sheet and its prompt are free on their own listing.

**Is any of it copied?** No. We read 12 public build reports, threads and tools and wrote everything in our own words. Sources are linked and credited inside.

## How to buy

$29 one time at https://paperchasewebb.ai/bots/marketplace/items/agents-that-use-a-computer-approval-gates-and-status-signals/. Stripe checkout, then an instant download page; the download link is also in your emailed receipt.

Seller: PaperChaseWebb Inc., mgmt@paperchasewebb.com