LLeMbas Desktop 0.7.1, carrying LLeMbas CLI 1.5.1

The microphone: WebKitGTK refuses a permission request its app does not
answer, and has no prompt of its own, so dictation on the server's page was
always blocked. The content view now allows the microphone -- only that, and
only there -- and media capture is switched on explicitly for the older
WebKitGTK the AppImage carries.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

CHANGELOG.md

@@ -6,6 +6,24 @@ All notable changes to LLeMbas Desktop. The format follows
 
 ## [Unreleased]
 
+## [0.7.1] — 2026-10-23
+
+The microphone works, and LLeMbas CLI 1.5.1 inside.
+
+### Fixed
+
+- **Dictation said the microphone was blocked, with no way to allow it.** The server's page asks
+  for the microphone, and a webview grants it only if the app answers: WebKitGTK refuses whatever
+  it is not answered on, and shows no prompt of its own. Desktop now allows the microphone -- the
+  microphone only, and only for the server's page in the content view; a camera or a screen still
+  gets the platform's refusal. Media capture is also switched on explicitly, for the older
+  WebKitGTK the AppImage carries.
+
+### Changed
+
+- Carries LLeMbas CLI 1.5.1: an image counts by its size, so a screenshot no longer makes the
+  agent compact.
+
 ## [0.7.0] — 2026-10-20
 
 Carries LLeMbas CLI 1.5.0: harness spec 2.0.0 — background work as jobs the person can see,

lembas-cli.version

@@ -1 +1 @@
-1.5.0
+1.5.1

src-tauri/Cargo.lock

@@ -2064,7 +2064,7 @@ checksum = "47d9d19d1d6efa0109d2f65ff4c85cddd50bd572e5a00127ab10987290bcefae"
 
 [[package]]
 name = "llembas-desktop"
-version = "0.7.0"
+version = "0.7.1"
 dependencies = [
  "gtk",
  "libc",

src-tauri/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "llembas-desktop"
-version = "0.7.0"
+version = "0.7.1"
 description = "LLeMbas Desktop: your LLeMbas server in a window, with the agent on this computer"
 authors = ["Jaroslav Beneš"]
 license = "MIT"
@@ -17,7 +17,7 @@ crate-type = ["staticlib", "cdylib", "rlib"]
 tauri-build = { version = "2", features = [] }
 
 [dependencies]
-tauri = { version = "2", features = ["tray-icon", "unstable"] }
+tauri = { version = "2.12", features = ["tray-icon", "unstable"] }
 tauri-plugin-dialog = "2"
 tauri-plugin-opener = "2"
 tauri-plugin-single-instance = "2"

src-tauri/src/lib.rs

@@ -30,7 +30,7 @@ use std::sync::Mutex;
 use serde::Serialize;
 use tauri::menu::{Menu, MenuItem, PredefinedMenuItem};
 use tauri::tray::TrayIconBuilder;
-use tauri::webview::WebviewBuilder;
+use tauri::webview::{PermissionKind, PermissionResponse, WebviewBuilder};
 use tauri::window::{Color, WindowBuilder};
 use tauri::{
     AppHandle, Emitter, LogicalPosition, LogicalSize, Manager, RunEvent, Url, Webview, WebviewUrl,
@@ -425,22 +425,56 @@ fn build_main(handle: &AppHandle, target: Option<Url>) -> tauri::Result<()> {
         pin_titlebar(&bar);
     }
     let nav = handle.clone();
+    let perm = handle.clone();
     window.add_child(
         WebviewBuilder::new(CONTENT, url)
             .user_agent(&user_agent())
             .background_color(theme.background())
             .initialization_script(THEME_SCRIPT)
-            .on_navigation(move |url| allowed(&nav, url)),
+            .on_navigation(move |url| allowed(&nav, url))
+            .on_permission_request(move |view, kind| permission(&perm, &view, kind)),
         LogicalPosition::new(FRAME, FRAME + TITLE_H),
         LogicalSize::new(
             size.width - 2.0 * FRAME,
             (size.height - TITLE_H - 2.0 * FRAME).max(0.0),
         ),
     )?;
+    #[cfg(target_os = "linux")]
+    if let Some(content) = handle.get_webview(CONTENT) {
+        enable_capture(&content);
+    }
     layout(&window);
     Ok(())
 }
 
+/// The server's dictation asks for the microphone, and a webview answers a permission request
+/// only if its app does: WebKitGTK refuses anything it is not answered on, and has no prompt of
+/// its own -- so the microphone simply said "Permission may be blocked" with no way to allow it.
+/// Granted for the microphone, in the content view, once a server is chosen, and for nothing
+/// else (a camera or a screen gets the platform's default, which on Linux is no). The content view
+/// never shows another origin: `allowed` sends every other site to the system browser.
+fn permission(handle: &AppHandle, view: &Webview, kind: PermissionKind) -> PermissionResponse {
+    if kind != PermissionKind::Microphone || view.label() != CONTENT {
+        return PermissionResponse::Default;
+    }
+    if handle.state::<App>().settings().server.is_some() {
+        PermissionResponse::Allow
+    } else {
+        PermissionResponse::Deny
+    }
+}
+
+/// Media capture on, explicitly: current WebKitGTK has it on by default, but the AppImage carries
+/// the older WebKitGTK of the system it was built on.
+#[cfg(target_os = "linux")]
+fn enable_capture(view: &Webview) {
+    let _ = view.with_webview(|pv| {
+        use gtk::prelude::*;
+        let settings: gtk::glib::Object = pv.inner().property("settings");
+        settings.set_property("enable-media-stream", true);
+    });
+}
+
 /// On Linux, Tauri packs every webview of a window into the window's GTK box, each with an equal
 /// share of the height, and positions and sizes set on a webview are not applied there. So the
 /// title bar is told directly: a fixed height, no share of what is left (the content takes it).