DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to approve risky agent commands from your iPhone with Brew Ping

Hold dangerous commands from a coding agent until you approve or deny them on your phone.

Credentials. The paired token is equivalent to command execution on that computer, so protect ~/.brewping/pairing.json and the paired phone.

Who it is forA developer who starts a coding agent on a Mac or Windows computer and leaves the desk while it works.

The problemA coding agent may try a risky shell command while nobody is at the machine to check it.

The outcomeRisky commands wait for your approval on the phone, and requests you do not answer time out as denied.

What you need first

  • A Mac or Windows computer with BrewPing Desktop installed (macOS 13.0+ or Windows 10/11 with WebView2)
  • At least one CLI coding agent installed on that computer, such as Claude Code, Codex CLI, OpenCode, or pi
  • An iPhone (iOS 17.0+) with the BrewPing app from the App Store, or an Android build from the repository
  • Phone and computer on the same local network, since cross-network access is not supported

Steps

  1. Download Brew Ping Desktop from the GitHub Release page at https://github.com/banmu123/BrewPing/releases/tag/v1.0.0 and install it.
  2. Install the agent you want to use, for example npm install -g @anthropic-ai/claude-code.
  3. In Brew Ping Desktop, open Show Pairing Code, then scan the QR code on your iPhone or tap Add Device and type the 6-digit code.
  4. Pick the agent, choose a working folder, and set the approval mode to safe or ask-all.
  5. Send an instruction that may run a shell command, then approve or deny the request on the phone.

Prompt for your agent

Using the repo banmu123/BrewPing, confirm that BrewPing Desktop lists the agents installed on this computer and that the approval mode for <target folder> is safe. Expected output: the detected agents and the active approval mode. Verify that a pending risky command shows as waiting and does not run until it is approved.

Caveat. The approval gate covers commands as they enter the agent, not shell commands the agent spawns on its own, so it is a safety net and not a sandbox.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for banmu123/BrewPing and "Approve risky agent commands from your iPhone"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer