I run several Claude Code and OpenCode sessions at once, often across 2+ devboxes and my local machine. I wanted a way to talk to my Claude Code session without having to be in the window/terminal. I wanted a way to see all my tmux sessions across machines from a single interface. So, I built agentwire-dev.
AgentWire is a self-hosted web portal and CLI that sits on top of tmux. It gives you a single dashboard where you can see, manage, and talk to all your coding agent sessions — local or remote. You push-to-talk from your browser (phone, tablet, laptop), it transcribes and routes to whichever session you pick, and the agent talks back via TTS. Think of it as a walkie-talkie for your AI coding setup.
What it does:
- *Push-to-talk from any device* — browser-based portal, works on your phone across your local network
- *Multi-session management* — create, switch between, and monitor tmux sessions running Claude Code, OpenCode, or any CLI agent
- *Worker pane orchestration* — spawn worker panes inside a session, send them tasks, get summaries back
- *TTS/STT* — agents speak their responses, you speak your prompts (Whisper for STT, Chatterbox or Qwen3-TTS for voices)
- *Git worktrees* — parallel workers can commit to separate branches without conflicts
- *Safety hooks* — pre/post hooks on tool calls for guardrails
- *Multi-machine* — SSH into remote boxes and manage sessions there too
It's `pip install agentwire-dev` and self-hosted. No cloud, no accounts. Your voice audio stays on your machine.
I use this every day. It's rough — the portal UI needs work, error handling is spotty in places, and the docs are thin. But the core workflow (talk to agents, they talk back, orchestrate multiple sessions) is solid for my setup.
Looking for feedback on:
- *Voice UX* — Is push-to-talk the right interaction model? Or would wake-word / continuous listening be better for coding workflows?
- *Multi-agent orchestration* — If you run multiple AI coding sessions, what would you actually want from a coordination layer?
- *What's missing* — What would make this useful for your workflow?
Built with Python, tmux, MCP protocol. Works with Claude Code and OpenCode out of the box, but anything that runs in a terminal can be a session.
I run several Claude Code and OpenCode sessions at once, often across 2+ devboxes and my local machine. I wanted a way to talk to my Claude Code session without having to be in the window/terminal. I wanted a way to see all my tmux sessions across machines from a single interface. So, I built agentwire-dev.
AgentWire is a self-hosted web portal and CLI that sits on top of tmux. It gives you a single dashboard where you can see, manage, and talk to all your coding agent sessions — local or remote. You push-to-talk from your browser (phone, tablet, laptop), it transcribes and routes to whichever session you pick, and the agent talks back via TTS. Think of it as a walkie-talkie for your AI coding setup.
What it does:
- *Push-to-talk from any device* — browser-based portal, works on your phone across your local network - *Multi-session management* — create, switch between, and monitor tmux sessions running Claude Code, OpenCode, or any CLI agent - *Worker pane orchestration* — spawn worker panes inside a session, send them tasks, get summaries back - *TTS/STT* — agents speak their responses, you speak your prompts (Whisper for STT, Chatterbox or Qwen3-TTS for voices) - *Git worktrees* — parallel workers can commit to separate branches without conflicts - *Safety hooks* — pre/post hooks on tool calls for guardrails - *Multi-machine* — SSH into remote boxes and manage sessions there too
It's `pip install agentwire-dev` and self-hosted. No cloud, no accounts. Your voice audio stays on your machine.
I use this every day. It's rough — the portal UI needs work, error handling is spotty in places, and the docs are thin. But the core workflow (talk to agents, they talk back, orchestrate multiple sessions) is solid for my setup.
Looking for feedback on:
- *Voice UX* — Is push-to-talk the right interaction model? Or would wake-word / continuous listening be better for coding workflows? - *Multi-agent orchestration* — If you run multiple AI coding sessions, what would you actually want from a coordination layer? - *What's missing* — What would make this useful for your workflow?
Built with Python, tmux, MCP protocol. Works with Claude Code and OpenCode out of the box, but anything that runs in a terminal can be a session.