Agent Commander

Agent Commander (AgentCmd) — drive the coding agent on your computer from your phone. Hand a task to the coding agent on your computer from wherever you are, watch every step as it happens, answer when it asks, approve its plan, and pick up the files it produces. The app talks only to Agent Commander, installed on your own computer next to the agent you already use. Nothing runs on our servers, and there is no account to create. BEFORE YOU START • Install Agent Commander on the computer where Claude Code CLI or Codex CLI is set up. The installer prints a QR code; scan it with the app and you are paired. Setup guide: agentcmd.app/support • The coding agent itself is not included. The app drives the copy already installed on your machine. • Works with Claude Code CLI and Codex CLI. Antigravity is an experimental option once its CLI is installed and signed in on the same computer. WHAT YOU CAN DO • Work from anywhere — Type a request, attach a photo or a file, and the agent runs it inside the project folder you chose. The transcript streams to your phone as it happens: every tool call, every file it reads or edits, every diff. Type @ for a file, / for a skill. • Stay in the loop, not in the chair — When the agent needs a decision, a card appears with a countdown: allow a tool, answer a question, approve a plan. Steer a running turn with an extra message, or queue up the next one. A turn can run for many minutes; close the app and a notification brings you back when it finishes or needs you. • Pick the level of control — Default, Accept edits, Plan only, or Bypass permissions. Set it per project, change it mid-conversation, and save allow or deny rules for the tools you trust. • Pick up what it made — Files the agent produced sit in one place. Preview images, Markdown, JSON and text right in the app, then share them. • Browse and edit your project — Open the file tree, read with syntax highlighting, edit and save, create, rename, move or delete. Git badges show what changed, and any file's commit history is one tap away. • Review the work — A branch graph of your history, the diff of every commit and every file, pull, push, publish a branch. • Catch up on the terminal — Sessions you started with the CLI on that computer appear read-only, live while they run. Continue any of them in the app. • Keep a task list — Note what needs doing in each project, attach files, add comments, and link a task to the conversation that handles it. Tasks live on your computer, so they survive a reinstall. • Watch the meter — Tokens per turn, today's total, a per-project summary, and how much of your plan's 5-hour and weekly windows is left. • Configure projects in depth — Working directory and extra folders, system prompt, permission rules, MCP servers, agents and skills. • Rewind and fork — Restore files to how they were before a turn, or fork a conversation to try another way. A context meter shows how full the window is; compact it yourself before the agent has to. • Several computers, one app — Pair each computer you code on, name it, and switch from the title bar. A notification opens the right one. • Languages — English and Vietnamese. Works on phones and tablets. BUILT FOR PRIVACY • No account, no sign-up, no profile. Pairing with your own computer is the only identity there is. • Your conversations, code and files stay on your computer, shared only with the AI provider of the agent you use. We have no copy or access. • Every connection is encrypted and pinned to your computer's certificate. • No ads, no in-app purchases, no tracking, no usage analytics. Anonymous crash reports only. • Remove a device at any time: the app wipes its local data on the spot. Agent Commander is an independent, third-party client. It is not affiliated with, endorsed by, or sponsored by Anthropic. Claude and Claude Code are trademarks of Anthropic. Privacy Policy: agentcmd.app/privacy-policy Support: agentcmd.app/support