feat(llm): add generate_tags command

Reuses build_prompt's transcript+metadata assembly (same as
generate_summary) and the same recording-in-progress guard.

src-tauri/src/commands.rs

@@ -1941,6 +1941,46 @@ pub async fn confirm_action_items(
     Ok(())
 }
 
+/// Suggests 1-8 short topical tags from the transcript (T8.3, FR-SEARCH-2) —
+/// same LLM path/prompt assembly as generate_summary, just a much shorter
+/// non-streamed reply. Suggestions are NOT saved automatically; the caller
+/// reviews/merges them and still calls set_tags to persist, same as a
+/// manually typed tag.
+#[tauri::command]
+pub async fn generate_tags(
+    state: State<'_, AppState>,
+    meeting_id: MeetingId,
+) -> WaResult<Vec<String>> {
+    let guard = state.session.lock().await;
+    if guard.as_ref().is_some_and(|s| s.meeting_id == meeting_id) {
+        return Err(WaError::new(
+            "llm",
+            "cannot generate tags while this meeting is still recording — wait until it's stopped",
+        ));
+    }
+    drop(guard);
+
+    let settings = load_settings();
+    let provider = llm_provider_from_settings(&settings).ok_or_else(|| {
+        WaError::new(
+            "llm",
+            "no LLM provider is configured — enable one in Settings first",
+        )
+    })?;
+
+    let meeting = state
+        .store
+        .get_meeting(&meeting_id)
+        .await
+        .map_err(|e| WaError::new("storage", e.to_string()))?;
+    let prompt = build_prompt(&meeting, None);
+    let text = format!("{}\n\n{}", prompt.metadata, prompt.transcript);
+    provider
+        .suggest_tags(&text)
+        .await
+        .map_err(|e| WaError::new("llm", e.to_string()))
+}
+
 // ---- Calendar / .pst (Phase 6) ----
 
 /// Import events + attendees from a `.pst` (T6.1/T6.2, FR-CAL-1). The