Agents At Work: You can Go!

About this app

Your AI agents are working. You don't have to watch. Agents At Work is the mobile companion for developers running Claude, Codex, Gemini, Grok CLI, Ollama (chat or agentic) on their Mac. Get instant push notifications when your agent needs approval, finishes a task, or hits an error, and respond from anywhere, without touching your Mac. WHY NOT JUST USE SOMETHING ELSE? - Claude.ai mobile lets you dispatch tasks to Claude while your Mac is awake: good for simple jobs. But it only controls Claude, offers limited real-time visibility, simpler permission prompts, and no rich conversation feed. No Codex, Grok, or Gemini support. - Generic push tools (Pushover, ntfy, etc.) only send notifications with zero context or ability to reply/approve. - OpenClaw provides remote control via Telegram for quick commands, but it requires complex setup and lacks a dedicated rich conversation UI with full context. Agents At Work is different. WHAT YOU GET Push Notifications: Your phone buzzes the moment your agent needs input or completes work. Rich notifications show what happened so you can act without opening the app. Permission Prompt Approvals: When your agent asks to run a command, edit a file, or use a tool, the prompt lands on your phone. Tap Yes, No, or "Yes for this session"; your agent continues instantly. Auto-Approve Mode: Enable auto-approve with one tap for trusted operations. Your agent runs without interruption; you stay in the loop without lifting a finger. Conversation Thread View: Every message, every tool call, every response in a clean scrollable feed. Tap to expand long output. Exactly what's on your Mac screen, in real time. Remote Messaging: Send follow-up instructions from your phone. Type from anywhere; your agent receives it instantly. Multi-Agent Support: Running Claude Code on one project and Codex on another? Both appear in Agents At Work, color-coded and labeled. Claude is purple, Codex is teal, Gemini is Google blue, Grok is black, Ollama is yellow. Session Status: Live status for every session: Running, Thinking, Waiting for Input, or Finished. End-to-End Encrypted: Your prompts and agent responses are encrypted on your Mac with AES-256 before they leave for the cloud. The encryption key is exchanged directly between your Mac and your phone via a one-time QR scan. Firebase never sees your conversation content in plaintext. Your code stays on your machine. Your conversations stay between your devices. HOW IT WORKS 1. Download Agents At Work on your iPhone. 2. Install the free macOS companion on your Mac (download at agentsatwork.app). 3. Sign in on your Mac with Google or email/password. 4. In the Mac app, click Link Mobile; a QR code appears. On your iPhone, open Agents At Work and tap Scan QR Code. That one scan links your phone and sets up end-to-end encryption. No separate account needed on your phone. 5. Launch any Claude Code, Codex, Gemini, Grok, or Ollama session as usual. 6. Your phone gets notified. Respond, approve, or monitor from anywhere. Your code, API keys, and agent output never leave your Mac. Events flow to your phone through your own secure Firebase account, encrypted end-to-end. We don't have a backend that reads your code. SUPPORTED AGENTS • Claude Code by Anthropic • OpenAI Codex CLI • Google Gemini CLI • Grok CLI • Ollama REQUIREMENTS • iPhone with iOS 17 or later • Mac with macOS 13 Ventura or later (companion app required) • Claude Code, Codex CLI, Gemini CLI, or Grok CLI installed on your Mac, or Ollama PRIVACY Agents At Work does not collect or store your code, prompts, or agent responses on any server we control. Conversation content is encrypted end-to-end on your Mac before it reaches Firebase, only your devices hold the key. We collect only what is necessary to deliver notifications and manage your subscription. Put your Mac to work. Take your phone anywhere. Terms of Use: https://www.apple.com/legal/internet-services/itunes/dev/stdeula/

Ollama is here. Run any open-source AI model on your Mac and control it from your phone, the same way you would Claude, Codex, Gemini, or Grok. - Any model works for chat and writing - Models in the `qwen2.5` and `qwen3` families unlock Agentic mode: the model reads files, writes code, and runs shell commands on your Mac while you are away - Before doing anything potentially destructive, it pauses and sends you a permission prompt to approve or deny right on your phone Mac session window: - Mobile/Desktop indicator and auto-approve badge now visible in the status bar - Press up and down arrow keys in the message input to cycle through your last 10 sent prompts

Momentum comes from crawler snapshots for the tracked storefront. “Viral growth signal” means unusually fast recent rating gains; it is an indicator, not download data.

1. First observed Aug 16, 2026 · 8:53 PM Free