Jazzen Chen

VibeAround v0.7.17 is out.

This release focuses on a deceptively hard problem: when the same local coding-agent session is accessible from Web, TUI, and messaging channels, every surface needs to agree on what active, queued, stopped, and finished mean.

VibeAround started as a desktop hub for your local AI coding agents. With v0.7.10 it's no longer shaped only as a desktop app: the local server, launch layer, Desktop UI, Web Hub, CLI, and TUI are now separate pieces that can run and evolve independently so VibeAround can meet you from whatever surface you actually work in.

Highlights

• Standalone server the VibeAround runtime now runs as a vibearound-server binary without the Desktop UI, still serving the Web Hub, local API bridge, MCP endpoints, previews, pairing, and sessions. Desktop is now one client of the runtime, not the runtime itself.

• CLI + one-line install install the VibeAround CLI from npm, then run `va serve` to start the server. It also includes commands for launch profiles, sessions, workspaces, previews, diagnostics, and chat workflows.

• Unified launch layer (va-launch) agent launching is now a standalone boundary with explicit launch profiles, so the same agent + profile + workspace + session launches reproducibly from Desktop, CLI, scripts, or automation without touching each agent's own config.

• Early terminal hub (va-tui) a terminal-first alternative to the Web Hub: connect to local sessions, switch agent context, and chat from your terminal. Early but moving fast.

I ve been shipping VibeAround pretty actively recently, and v0.7.7 is one of the bigger workflow updates so far.

VibeAround is an all-in-one hub for AI coding agents. The goal is to keep the agents you already use, such as Claude Code, Codex CLI, Gemini CLI, OpenCode, Claude Desktop, and Codex Desktop, in one workflow: launch, bridge, inspect, continue, preview, and control them across devices.