FastAF runs all your Claude Code and Codex sessions in one window and lets you talk to them. A voice agent writes the prompt, sends it to the right terminal, waits for the agent to finish and reads the result back. Speech in and out runs on your machine (Whisper + Kokoro). See which agent is working, waiting or done, tile up to 9 in Multiview, and review diffs and PRs. Native Rust + Tauri, ~300 MB idle. Free and open source (Apache 2.0) for macOS, Windows and Linux.

Hey Product Hunt! I'm Karam, and FastAF is my first app. I run four or five Claude Code and Codex sessions at once, and I kept missing the one stuck on a permission prompt or a rate limit. So I built a workspace that watches every terminal and lets me just talk to them. How it works: you speak, on-device Whisper transcribes, a model you pick (local Ollama, or Anthropic/OpenAI/OpenRouter) writes a precise prompt and sends it to the right terminal, then it waits for the agent to go idle and tells you out loud what happened. You can interrupt it mid-sentence. It's free and open source (Apache 2.0), native Rust + Tauri, and it started as a fork of TUICommander by sstraus. I'd love feedback on the voice flow: what would make you talk to your agents instead of typing?

FastAF

Free, open-source voice workspace for Claude Code and Codex

Free, open-source voice workspace for Claude Code and Codex