dsh-agent-voice 0.1.6 → 0.1.7

package/CHANGELOG.md

@@ -1,3 +1,8 @@
+## 0.1.7
+
+- System-level playback: announcement WAV is played through macOS afplay (independent of browser autoplay policy), so the voice is heard on every tab and in the background.
+- macOS notification (osascript) on attention requests: visible alert + sound when the agent needs the user at the keyboard.
+
 # Changelog
 
 ## 0.1.0

package/lib/index.js

@@ -125,6 +125,34 @@ function apply(ctx) {
     return await readFile(res.audio);
   }
 
+  // System-level playback: browsers block audio.play() until a user gesture,
+  // so the in-page overlay player alone is unreliable. Playing the WAV through
+  // macOS afplay guarantees the announcement is heard on every tab (and even
+  // when the browser is in the background). Fire-and-forget.
+  async function systemPlay(wav) {
+    try {
+      const tmp = join(AUDIO_DIR, "tmp-" + String(Date.now()) + "-" + Math.random().toString(36).slice(2) + ".wav");
+      await writeFile(tmp, wav);
+      const af = spawn("/usr/bin/afplay", [tmp], { detached: true, stdio: "ignore" });
+      af.unref?.();
+      setTimeout(() => {
+        try { spawn("/bin/rm", ["-f", tmp]); } catch {}
+      }, 30000).unref?.();
+    } catch {}
+  }
+
+  // macOS notification (attention only): visible alert + sound even when the
+  // browser window is not focused — the "I need you at the keyboard" signal.
+  function systemNotify(title, text) {
+    try {
+      const scr = [
+        'display notification ' + JSON.stringify(String(text).slice(0, 120)) + ' with title ' + JSON.stringify(title),
+      ].join("\n");
+      const osa = spawn("/usr/bin/osascript", ["-e", scr], { detached: true, stdio: "ignore" });
+      osa.unref?.();
+    } catch {}
+  }
+
   async function announce(kind, text) {
     const cfg = source();
     if (!cfg.enabled) return { ok: false, skipped: "disabled" };
@@ -141,6 +169,11 @@ function apply(ctx) {
     await ctx.settings.replace(STATUS_NS, { json: JSON.stringify(payload) });
     audioCache.set(id, wav);
     setTimeout(() => audioCache.delete(id), 60000).unref?.();
+    // Always play system-wide (independent of browser autoplay policy).
+    systemPlay(wav);
+    if (kind === "attention") {
+      systemNotify("🔔 DSH — potřebuji tě", text);
+    }
     return { ok: true, kind, text };
   }
 

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "dsh-agent-voice",
-  "version": "0.1.6",
-  "description": "DeepSeek Harness plugin: gives the agent a voice — it reads completed tasks and attention requests aloud (never reasoning or interim reports) via local Piper / XTTS v2, with a settings panel and a curated set of voices.",
+  "version": "0.1.7",
+  "description": "DeepSeek Harness plugin: gives the agent a voice \u2014 it reads completed tasks and attention requests aloud (never reasoning or interim reports) via local Piper / XTTS v2, with a settings panel and a curated set of voices.",
   "type": "module",
   "main": "lib/index.js",
   "files": [
@@ -17,7 +17,7 @@
     "node": ">=22"
   },
   "license": "MIT",
-  "author": "Michal Ožuch",
+  "author": "Michal O\u017euch",
   "keywords": [
     "deepseek-harness",
     "dsh",
@@ -57,4 +57,4 @@
     "@deepseek-ai/dsh-settings": "^0.1.1-rc.2"
   },
   "dependencies": {}
-}
+}