Agents at work.
Life goes on.

Agents run in tmux. Disconnect, close your laptop and attach again where you left off.

Provision MCP servers, shared configuration, skills and plugins. The session image includes headless Chromium/Playwright and code tooling.

Browse a persistent project and conversation catalog without waking sessions. Explicitly continue in the original environment. Requires the custom server build; shell !commands are not routed.

Enable a separate critic to review an agent turn before completion is announced. Inspect its findings and review outcomes.

Change the harness or switch accounts in a running session, with conversation resumption where supported.

Configure Claude and Codex accounts, project pinning, preferences, quota thresholds and rotation. Opt-in; provider terms apply.

Start, attach, name, stop, resume and end sessions. Use scratch workspaces, recent targets, host selection, zsh completion and dev doctor.

Discover GitHub and Gitea repositories, browse project folders and follow clone progress. Inspect commits, push branches or push and create a pull request.

Paste a Mac screenshot into Claude Code, Codex or Grok. The clipboard bridge delivers the image to the remote agent.

Copy a file in Finder and paste it into the session’s shared /inbox. Larger files can go through the SMB mount.

Get notified when an agent finishes, needs an answer or permission, or gets stuck. Click through to the relevant Warp tab.

See working, waiting and done states in terminal tabs. Restore exact sessions after Warp or macOS restarts.

Install the listener as a LaunchAgent, configure path mappings and use dev doctor to check the connection.

Optional Xcode MCP and mac-sync bridges let remote agents work with Xcode on your Mac. An OpenDesign bridge connects to its installed Mac runtime. Operations require the target Mac online and approved SSH.

Optionally include local Claude usage in the same statistics, with host filtering and project attribution where available.

Find branches and sessions with local changes or commits that have not reached the remote.

Give a guest revocable, time-limited access to a live terminal and development-server previews, without access to your tailnet.

Review notifications, activity, agent upkeep, account settings and host health without changing tools. Inspect notification decision history and send feedback on wrong or missing alerts.

Inspect quota windows and reset times, connect provider accounts, configure models and context windows, and edit service settings.

Inspect CPU, RAM, disk, network, uptime and backup health. Manual backup, update and reboot actions include confirmation flows.

Use APNs or Pushover for phone notifications. Mark a session important when it needs a louder alert.

Connect over Tailscale, or configure Cloudflare Tunnel and Access for access outside the tailnet.

Browse overview, statistics, AI limits, hardware, folders and activity. Inspect terminal/session actions and run, toggle or delete schedules; create and edit schedules on the web.

Route production database access through a host proxy, with a separate approval step for writes.

The control plane binds to the Tailscale network. Optional Cloudflare Access provides another authenticated route.

Read the security model and deployment guidance before connecting sensitive systems or credentials.

Configure read, DML and migration scopes. Preview SQL impact without executing it, approve exact writes and inspect their audit/results. PostgreSQL and MariaDB have engine-specific setup paths.

Prepare access plans, inspect preflight checks and approve resumable setup operations while database credentials remain host-side.

Run configured Coolify deployments, optionally pin an exact commit, or sequence a deploy after an approved successful database write. Supported application restarts require approval.

Use the included automation to deploy main after successful CI. Configure it for your own installation.

Set per-session memory limits and inspect machines, storage and resource history in the dashboard.

Borg backups and CLI backup/restore support recovery of volumes, secrets and the database.

Recovery-verified storage cleanup and monitoring help keep the host maintainable.

Optional external monitoring and a watchdog can detect control-plane outages. Integrates with Beszel and Uptime Kuma.

The optional CI deployment workflow includes post-deploy health checks and rollback. Configure monitoring and backup targets for your own host.

Optional memory/OOM guards, disk budgets and cleanup that requires verified offsite archives support recovery-conscious maintenance.