Skip to content

v0.5.0 — Voice input

Latest

Choose a tag to compare

@github-actions github-actions released this 18 Jun 11:26
· 3 commits to main since this release
2ef0928

FlyCrys v0.5.0

✨ New

Voice input — Click the 🎤 in the chat toolbar and dictate your message; it streams into the input as you speak. Powered by Claude CLI's built-in speech-to-text. Requires a Claude.ai login and a system audio recorder (PipeWire/PulseAudio, ALSA, SoX, or ffmpeg).

Model & effort switcher — Choose the model and reasoning effort straight from the chat toolbar. The list is fetched live from the CLI (nothing hardcoded), the model version is shown in the selector, and the active model · effort appears in the status bar before your first message.

⚡ Performance

  • File viewer never blocks the UI — syntax highlighting runs off the main thread and paints in idle chunks, so large or minified files open instantly.
  • Instant tab switching — workspaces build progressively and chat history parses off-thread.
  • No more main-thread stalls — git status and autosave moved to worker threads.
  • Stick-to-bottom chat scroll — stays out of your way when you scroll up to read back while the agent is working.

🛠 Other

  • "None of these" escape hatch on AskUserQuestion cards.
  • Desktop notification when the agent asks a question (quotes the workspace name).

Full changelog: v0.4.0...v0.5.0