Sound Feedback Validation

Sound feedback lets you keep track of builds, deploys, commits, and the chat by ear — without watching the screen. Xedant Agent plays a distinct sound for each event, so you can start a long task in the background and get on with other things: the sound alone tells you whether everything is going well or it is time to come back to the screen.

---

Built-in Sounds

Xedant Agent ships with six built-in sounds, and each one has its own event:

• message — for every new AI response: it confirms the model is still working
• complete — when a chat or a background task finishes processing
• notification — when the model is waiting for your decision or confirmation (for example, it asked you a question)
• error — when a build or a task fails
• commit — when changes are saved to git
• deploy — when a deploy starts or finishes

Over time you will start to read the work by ear: a steady rhythm of messages means everything is going smoothly, while a sudden silence, a notification chime, or an error sound is your cue to look at the screen.

---

Build and Deploy Sounds

Each build can have its own sounds that override the shared ones. Five sound fields are supported: startSound plays when the build starts, and four result fields decide what plays when it finishes. The outcome-specific sounds — successSound (success), warningSound (warning), errorSound (error) — take priority over the general endSound, so you can assign a different sound to each outcome. For example, give a critical security check a loud error sound when it fails, and the tests a success tune when they pass.

build:
  security-scan:
    command: "npm run security-audit"
    watch: "*.ts"
    startSound: "notification"
    successSound: "complete"
    errorSound: "error"
    soundVolume: 1.0

  lint:
    command: "npm run lint"
    watch: "*.ts"
    startSound: ""
    endSound: ""

If a sound field is empty, no sound plays for that event. endSound is the universal fallback: it plays whenever the build’s result has no sound of its own. The soundVolume field (0 to 1, default 1.0) sets the volume independently of the general settings.

---

Sound Settings

There are two ways to control the sounds:

• Mute everything — the speaker icon in the chat header instantly turns all sounds on or off. The fastest way to silence everything at once
• Volume per type — in Settings → Sound, every sound type has its own volume slider (0–100%). You can turn noisy sounds down while keeping the important ones loud

All sound settings sync between browser tabs in real time: mute the sound in one tab, and it goes silent everywhere.

---

Custom Sounds

You can replace any built-in sound: put an MP3 file with the same name into the .xedant/sounds/ folder of your project — for example, message.mp3, complete.mp3, error.mp3, commit.mp3, deploy.mp3. Your file takes priority over the built-in one.

Build and deploy sounds can use any name: the app first looks for .xedant/sounds/{name}.mp3, and if it is not there, falls back to the built-in wwwroot/sounds/{name}.mp3. This makes it easy to give different builds different sounds — just give them different names in the settings.

---

Practical Tips

• Use headphones — with Bluetooth headphones you can wait for tasks to finish without sitting in front of the screen: every sound tells you clearly what is going on
• Turn down noisy builds — builds that watch files and restart often will flood you with start and finish sounds. Set soundVolume: 0 for them or leave startSound empty, and save the sounds for the events that matter — a deploy finishing or a security check
• Silence is a cue to check — if you normally hear a steady stream of messages and it suddenly goes quiet, look at the screen: most likely the model has hit an error or is waiting for your answer
• Confetti on commits — every successful commit comes with a confetti animation and a sound. If it distracts you, switch off the Confetti On toggle in Settings → Sound

---