From 33d2d6be27a154722c717d59a72ff07521c126d6 Mon Sep 17 00:00:00 2001
From: krasi <krasimir.xristov@gmail.com>
Date: Sat, 16 May 2026 19:02:40 +0200
Subject: [PATCH 1/2] feat(audio): add rename-title endpoint and inline edit UI

---
 backend/app/features/audio/router.py          |  40 +++-
 backend/app/features/audio/schemas.py         |   6 +
 backend/app/features/audio/service.py         | 113 ++++++++++-
 frontend/app/(dashboard)/audio/page.tsx       |  17 +-
 .../features/audio/components/AudioCard.tsx   | 178 ++++++++++++++----
 frontend/features/audio/hooks/useAudio.ts     |  16 +-
 frontend/lib/api.ts                           |  10 +
 7 files changed, 336 insertions(+), 44 deletions(-)

diff --git a/backend/app/features/audio/router.py b/backend/app/features/audio/router.py
index 2eb791e..0f51151 100644
--- a/backend/app/features/audio/router.py
+++ b/backend/app/features/audio/router.py
@@ -1,20 +1,38 @@
-"""Audio Library REST endpoints — list and delete saved audio entries."""
+"""Audio Library REST endpoints — list, delete, and temp-serve audio."""
 
 import logging
 from uuid import UUID
 
 from fastapi import APIRouter, Depends, HTTPException, Request, status
+from fastapi.responses import Response
 
 from app.core.limiter import limiter
 from app.core.security import get_current_user
 from app.features.auth.schemas import UserSchema
 from app.features.audio import service
-from app.features.audio.schemas import AudioEntryOut
+from app.features.audio.schemas import AudioEntryOut, AudioEntryPatch
 
 logger = logging.getLogger(__name__)
 router = APIRouter(prefix="/audio", tags=["audio"])
 
 
+@router.get("/temp/{token}")
+async def serve_temp_audio(token: str) -> Response:
+    """Serve a temporarily stored mp3 before the user confirms saving.
+
+    No JWT auth — the UUID token itself acts as a short-lived capability URL
+    (unguessable, 1-hour TTL).  The browser <audio> element cannot send custom
+    headers, so cookie / bearer auth is not applicable here.
+    """
+    audio_bytes = service.retrieve_temp_audio(token)
+    if audio_bytes is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail="Audio not found or expired.",
+        )
+    return Response(content=audio_bytes, media_type="audio/mpeg")
+
+
 @router.get("", response_model=list[AudioEntryOut])
 @limiter.limit("60/minute")
 async def list_audio_entries_endpoint(
@@ -39,3 +57,21 @@ async def delete_audio_entry_endpoint(
             status_code=status.HTTP_404_NOT_FOUND,
             detail="Audio entry not found",
         )
+
+
+@router.patch("/{entry_id}", response_model=AudioEntryOut)
+@limiter.limit("60/minute")
+async def rename_audio_entry_endpoint(
+    entry_id: UUID,
+    body: AudioEntryPatch,
+    request: Request,
+    current_user: UserSchema = Depends(get_current_user),
+) -> AudioEntryOut:
+    """Rename a saved audio entry's title."""
+    updated = await service.rename_audio_entry(entry_id, current_user.id, body.title)
+    if updated is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail="Audio entry not found",
+        )
+    return updated
diff --git a/backend/app/features/audio/schemas.py b/backend/app/features/audio/schemas.py
index 4166abc..a2ded24 100644
--- a/backend/app/features/audio/schemas.py
+++ b/backend/app/features/audio/schemas.py
@@ -14,6 +14,12 @@ class AudioGenerateResult(BaseModel):
     text_preview: str = Field(..., min_length=1, max_length=1024)
 
 
+class AudioEntryPatch(BaseModel):
+    """Request body for PATCH /audio/{id} — rename the entry title."""
+
+    title: str = Field(..., min_length=1, max_length=200)
+
+
 class AudioEntryOut(BaseModel):
     """A saved audio entry from the audio_entries table."""
 
diff --git a/backend/app/features/audio/service.py b/backend/app/features/audio/service.py
index dda36fe..4df2d78 100644
--- a/backend/app/features/audio/service.py
+++ b/backend/app/features/audio/service.py
@@ -8,6 +8,16 @@
 generate_tts(text, user_id) -> tuple[bytes, str]
     Call OpenRouter TTS endpoint, return (mp3_bytes, storage_filename).
 
+store_temp_audio(audio_bytes) -> str
+    Write mp3 bytes to a system temp file and return a token (UUID).
+    TTL: 3600 s.  Survives uvicorn --reload (file system is persistent).
+
+retrieve_temp_audio(token) -> bytes | None
+    Return bytes for the given token, or None if missing / expired.
+
+delete_temp_audio(token) -> None
+    Remove an entry from the temp store (called after successful upload).
+
 upload_audio(audio_bytes, filename) -> str
     Upload mp3 bytes to Supabase Storage and return a signed URL (1 week).
 
@@ -23,13 +33,15 @@
 
 import asyncio
 import logging
+import tempfile
+import time
+from pathlib import Path
 from uuid import UUID, uuid4
 from typing import Any
 
 import httpx
 
 from app.core.config import settings
-from app.core.privacy import mask_pii
 from app.db.supabase import get_supabase_client
 from app.features.audio.schemas import AudioEntryOut
 
@@ -41,6 +53,41 @@
 # Signed URL expiry: 7 days in seconds.
 _SIGNED_URL_TTL = 7 * 24 * 60 * 60  # 604 800 s
 
+# ---------------------------------------------------------------------------
+# Temp audio store — bytes written to the system temp directory.
+# Survives uvicorn --reload (file system is persistent across restarts).
+# TTL enforced at retrieval time via file mtime.
+# ---------------------------------------------------------------------------
+
+_TEMP_DIR = Path(tempfile.gettempdir())
+_TEMP_PREFIX = "axon_audio_"
+_TEMP_TTL = 3600  # 1 hour
+
+
+def store_temp_audio(audio_bytes: bytes) -> str:
+    """Write mp3 bytes to a temp file and return a UUID token."""
+    token = str(uuid4())
+    path = _TEMP_DIR / f"{_TEMP_PREFIX}{token}.mp3"
+    path.write_bytes(audio_bytes)
+    return token
+
+
+def retrieve_temp_audio(token: str) -> bytes | None:
+    """Return mp3 bytes for the given token, or None if missing / expired."""
+    path = _TEMP_DIR / f"{_TEMP_PREFIX}{token}.mp3"
+    if not path.exists():
+        return None
+    if time.time() - path.stat().st_mtime > _TEMP_TTL:
+        path.unlink(missing_ok=True)
+        return None
+    return path.read_bytes()
+
+
+def delete_temp_audio(token: str) -> None:
+    """Delete the temp file after a successful Storage upload."""
+    path = _TEMP_DIR / f"{_TEMP_PREFIX}{token}.mp3"
+    path.unlink(missing_ok=True)
+
 
 async def generate_tts(text: str, user_id: str) -> tuple[bytes, str]:
     """Call OpenRouter TTS endpoint and return (mp3_bytes, storage_filename).
@@ -66,13 +113,10 @@ async def generate_tts(text: str, user_id: str) -> tuple[bytes, str]:
     if not text.strip():
         raise RuntimeError("Cannot generate TTS for empty text.")
 
-    # Mask PII before forwarding text to the external TTS provider.
-    masked_text = await mask_pii(text)
-
     # OpenRouter /audio/speech endpoint — returns raw MP3 bytes (not JSON).
     payload = {
         "model": settings.tts_model,
-        "input": masked_text,
+        "input": text,
         "voice": settings.tts_voice,
         "response_format": "mp3",
     }
@@ -345,3 +389,62 @@ async def delete_audio_entry(entry_id: UUID, user_id: str) -> bool:
 
     logger.info("[delete_audio_entry] deleted uuid=%s", safe_log_name)
     return True
+
+
+async def rename_audio_entry(
+    entry_id: UUID, user_id: str, title: str
+) -> AudioEntryOut | None:
+    """Rename an audio entry's title.
+
+    Returns the updated AudioEntryOut, or None if the entry was not found.
+    Raises RuntimeError on DB or Storage failure.
+    """
+    client = await get_supabase_client()
+
+    # UPDATE — postgrest-py's FilterRequestBuilder has no .select() after .eq(),
+    # so we do the update and re-fetch in a separate query.
+    try:
+        await (
+            client.table("audio_entries")
+            .update({"title": title})
+            .eq("id", str(entry_id))
+            .eq("user_id", user_id)
+            .execute()
+        )
+    except Exception:
+        logger.exception("[rename_audio_entry] UPDATE failed id=%s", entry_id)
+        raise RuntimeError("Failed to rename audio entry.") from None
+
+    # Re-fetch the updated row (also confirms the entry belongs to this user).
+    try:
+        fetch = (
+            await client.table("audio_entries")
+            .select(_SELECT_COLS)
+            .eq("id", str(entry_id))
+            .eq("user_id", user_id)
+            .execute()
+        )
+    except Exception:
+        logger.exception(
+            "[rename_audio_entry] SELECT after UPDATE failed id=%s", entry_id
+        )
+        raise RuntimeError("Failed to fetch renamed audio entry.") from None
+
+    if not fetch.data:
+        return None
+
+    row = fetch.data[0]
+    filename: str = row["filename"]
+    safe_log = filename.split("/")[-1] if "/" in filename else filename
+
+    try:
+        result = await client.storage.from_(settings.audio_bucket).create_signed_url(
+            filename, _SIGNED_URL_TTL
+        )
+        signed_url: str = result["signedURL"]
+    except Exception:
+        logger.exception("[rename_audio_entry] signed URL failed uuid=%s", safe_log)
+        raise RuntimeError("Title updated but failed to generate signed URL.") from None
+
+    logger.info("[rename_audio_entry] renamed uuid=%s", safe_log)
+    return _row_to_entry(row, signed_url)
diff --git a/frontend/app/(dashboard)/audio/page.tsx b/frontend/app/(dashboard)/audio/page.tsx
index 2b5c97c..a6f5678 100644
--- a/frontend/app/(dashboard)/audio/page.tsx
+++ b/frontend/app/(dashboard)/audio/page.tsx
@@ -1,12 +1,23 @@
+import Link from 'next/link';
+import { ArrowLeft } from 'lucide-react';
 import AudioLibrary from '@/features/audio/components/AudioLibrary';
 
 const AudioPage = () => {
   return (
     <div className='min-h-screen bg-[#13131b] px-4 py-8'>
       <div className='mx-auto max-w-5xl'>
-        <h1 className='mb-6 text-xl font-semibold text-[#e4e1ed]'>
-          Audio Library
-        </h1>
+        <div className='mb-6 flex items-center gap-3'>
+          <Link
+            href='/chat'
+            aria-label='Back to chat'
+            className='flex h-8 w-8 items-center justify-center rounded-md text-[#6b6b8a] transition-colors hover:bg-[#1e1e2e] hover:text-[#e4e1ed]'
+          >
+            <ArrowLeft className='h-5 w-5' />
+          </Link>
+          <h1 className='text-xl font-semibold text-[#e4e1ed]'>
+            Audio Library
+          </h1>
+        </div>
         <AudioLibrary />
       </div>
     </div>
diff --git a/frontend/features/audio/components/AudioCard.tsx b/frontend/features/audio/components/AudioCard.tsx
index ab1d154..67dd46f 100644
--- a/frontend/features/audio/components/AudioCard.tsx
+++ b/frontend/features/audio/components/AudioCard.tsx
@@ -1,10 +1,19 @@
 'use client';
 
-import { useState } from 'react';
+import { useState, useOptimistic, useTransition } from 'react';
 import { formatDistanceToNow } from 'date-fns';
-import { Trash2, Music } from 'lucide-react';
-import { Button } from '@/components/ui/button';
-import { useDeleteAudio } from '@/features/audio/hooks/useAudio';
+import { Trash2, Music, Pencil, Check, X } from 'lucide-react';
+import {
+  DialogRoot,
+  DialogPopup,
+  DialogTitle,
+  DialogDescription,
+  DialogClose,
+} from '@/components/ui/dialog';
+import {
+  useDeleteAudio,
+  useRenameAudio,
+} from '@/features/audio/hooks/useAudio';
 import type { AudioEntryOut } from '@/features/audio/types';
 
 interface AudioCardProps {
@@ -12,15 +21,51 @@ interface AudioCardProps {
 }
 
 const AudioCard = ({ entry }: AudioCardProps) => {
-  const [confirmDelete, setConfirmDelete] = useState(false);
-  const { mutate: deleteEntry, isPending } = useDeleteAudio();
+  const [deleteOpen, setDeleteOpen] = useState(false);
+  const [isEditing, setIsEditing] = useState(false);
+  const [draftTitle, setDraftTitle] = useState('');
+  const [, startTransition] = useTransition();
+  const [optimisticTitle, applyOptimisticTitle] = useOptimistic(
+    entry.title,
+    (_: string, newTitle: string) => newTitle,
+  );
+
+  const { mutate: deleteEntry, isPending: isDeleting } = useDeleteAudio();
+  const { mutateAsync: renameEntry } = useRenameAudio();
+
+  function startEdit() {
+    setDraftTitle(entry.title);
+    setIsEditing(true);
+  }
+
+  function cancelEdit() {
+    setIsEditing(false);
+  }
 
-  function handleDeleteClick() {
-    if (!confirmDelete) {
-      setConfirmDelete(true);
+  function saveEdit() {
+    const trimmed = draftTitle.trim();
+    if (!trimmed || trimmed === entry.title) {
+      setIsEditing(false);
       return;
     }
-    deleteEntry(entry.id);
+    setIsEditing(false);
+    startTransition(async () => {
+      applyOptimisticTitle(trimmed);
+      try {
+        await renameEntry({ id: entry.id, title: trimmed });
+      } catch {
+        // useOptimistic reverts to entry.title after the transition ends
+      }
+    });
+  }
+
+  function handleKeyDown(e: React.KeyboardEvent<HTMLInputElement>) {
+    if (e.key === 'Enter') saveEdit();
+    if (e.key === 'Escape') cancelEdit();
+  }
+
+  function confirmDelete() {
+    deleteEntry(entry.id, { onSuccess: () => setDeleteOpen(false) });
   }
 
   const relativeTime = formatDistanceToNow(new Date(entry.created_at), {
@@ -30,31 +75,66 @@ const AudioCard = ({ entry }: AudioCardProps) => {
   return (
     <div className='flex flex-col gap-3 rounded-lg border border-[#2a2a3d] bg-[#1a1a2e] p-4'>
       {/* Header row */}
-      <div className='flex items-start justify-between gap-2'>
-        <div className='flex min-w-0 items-center gap-2'>
-          <Music className='h-4 w-4 shrink-0 text-purple-400' />
+      <div className='flex items-center gap-2'>
+        <Music className='h-4 w-4 shrink-0 text-purple-400' />
+        {isEditing ? (
+          <input
+            autoFocus
+            maxLength={200}
+            value={draftTitle}
+            onChange={(e) => setDraftTitle(e.target.value)}
+            onKeyDown={handleKeyDown}
+            aria-label='Edit audio title'
+            className='min-w-0 flex-1 rounded border border-[#494bd6] bg-[#13131b] px-2 py-0.5 text-sm text-[#e4e1ed] outline-none'
+          />
+        ) : (
           <p
-            className='truncate text-sm font-medium text-[#e4e1ed]'
-            title={entry.title}
+            className='min-w-0 flex-1 truncate text-sm font-medium text-[#e4e1ed]'
+            title={optimisticTitle}
           >
-            {entry.title}
+            {optimisticTitle}
           </p>
-        </div>
-        <Button
-          size='icon'
-          variant='ghost'
-          onClick={handleDeleteClick}
-          disabled={isPending}
-          aria-label={confirmDelete ? 'Confirm delete' : 'Delete audio'}
-          className={
-            confirmDelete
-              ? 'shrink-0 text-red-400 hover:bg-red-900/30 hover:text-red-300'
-              : 'shrink-0 text-[#6b6b8a] hover:bg-[#2a2a3d] hover:text-[#e4e1ed]'
-          }
-          onBlur={() => setConfirmDelete(false)}
-        >
-          <Trash2 className='h-4 w-4' />
-        </Button>
+        )}
+        {isEditing ? (
+          <div className='flex shrink-0 items-center gap-0.5'>
+            <button
+              type='button'
+              onClick={saveEdit}
+              aria-label='Save title'
+              className='flex h-8 w-8 items-center justify-center rounded-md text-[#6b6b8a] transition-colors hover:bg-[#2a2a3d] hover:text-green-400'
+            >
+              <Check className='h-3.5 w-3.5' />
+            </button>
+            <button
+              type='button'
+              onClick={cancelEdit}
+              aria-label='Cancel edit'
+              className='flex h-8 w-8 items-center justify-center rounded-md text-[#6b6b8a] transition-colors hover:bg-[#2a2a3d] hover:text-[#e4e1ed]'
+            >
+              <X className='h-3.5 w-3.5' />
+            </button>
+          </div>
+        ) : (
+          <div className='flex shrink-0 items-center gap-0.5'>
+            <button
+              type='button'
+              onClick={startEdit}
+              aria-label='Edit title'
+              className='flex h-8 w-8 items-center cursor-pointer justify-center rounded-md text-[#6b6b8a] transition-colors hover:bg-[#2a2a3d] hover:text-[#c0c1ff]'
+            >
+              <Pencil className='h-3.5 w-3.5' />
+            </button>
+            <button
+              type='button'
+              onClick={() => setDeleteOpen(true)}
+              disabled={isDeleting}
+              aria-label='Delete audio'
+              className='flex h-8 w-8 items-center cursor-pointer justify-center rounded-md text-[#6b6b8a] transition-colors hover:bg-[#2a2a3d] hover:text-[#ff6b6b] disabled:cursor-not-allowed disabled:opacity-50'
+            >
+              <Trash2 className='h-4 w-4' />
+            </button>
+          </div>
+        )}
       </div>
 
       {/* Audio player */}
@@ -65,12 +145,44 @@ const AudioCard = ({ entry }: AudioCardProps) => {
         src={entry.signed_url}
         className='w-full rounded'
         preload='none'
-        aria-label={`Audio: ${entry.title}`}
+        aria-label={`Audio: ${optimisticTitle}`}
       />
       <p className='text-xs text-[#6b6b8a]/60 italic'>Captions unavailable</p>
 
       {/* Footer */}
       <p className='text-xs text-[#6b6b8a]'>{relativeTime}</p>
+
+      {/* Delete confirmation dialog */}
+      <DialogRoot open={deleteOpen} onOpenChange={setDeleteOpen}>
+        <DialogPopup>
+          <DialogTitle>Delete audio?</DialogTitle>
+          <DialogDescription>
+            This will permanently delete &ldquo;{entry.title}&rdquo; from your
+            library and remove the file from storage. This action cannot be
+            undone.
+          </DialogDescription>
+          <div className='mt-5 flex justify-end gap-2'>
+            <DialogClose
+              render={
+                <button
+                  type='button'
+                  className='cursor-pointer rounded-lg px-4 py-2 text-sm text-[#9b9bb8] transition-colors hover:bg-[#1e1e2e] hover:text-[#e4e1ed]'
+                />
+              }
+            >
+              Cancel
+            </DialogClose>
+            <button
+              type='button'
+              disabled={isDeleting}
+              onClick={confirmDelete}
+              className='cursor-pointer rounded-lg bg-[#ff4444]/10 px-4 py-2 text-sm text-[#ff6b6b] transition-colors hover:bg-[#ff4444]/20 disabled:cursor-not-allowed disabled:opacity-50'
+            >
+              {isDeleting ? 'Deleting…' : 'Delete'}
+            </button>
+          </div>
+        </DialogPopup>
+      </DialogRoot>
     </div>
   );
 };
diff --git a/frontend/features/audio/hooks/useAudio.ts b/frontend/features/audio/hooks/useAudio.ts
index 0198fb4..31537af 100644
--- a/frontend/features/audio/hooks/useAudio.ts
+++ b/frontend/features/audio/hooks/useAudio.ts
@@ -1,7 +1,11 @@
 'use client';
 
 import { useMutation, useQuery, useQueryClient } from '@tanstack/react-query';
-import { deleteAudioEntry, listAudioEntries } from '@/lib/api';
+import {
+  deleteAudioEntry,
+  listAudioEntries,
+  renameAudioEntry,
+} from '@/lib/api';
 import type { AudioEntryOut } from '@/features/audio/types';
 
 export const AUDIO_ENTRIES_KEY = ['audio-entries'] as const;
@@ -23,3 +27,13 @@ export function useDeleteAudio() {
     },
   });
 }
+
+export function useRenameAudio() {
+  const qc = useQueryClient();
+  return useMutation<AudioEntryOut, Error, { id: string; title: string }>({
+    mutationFn: ({ id, title }) => renameAudioEntry(id, title),
+    onSuccess: () => {
+      void qc.invalidateQueries({ queryKey: AUDIO_ENTRIES_KEY });
+    },
+  });
+}
diff --git a/frontend/lib/api.ts b/frontend/lib/api.ts
index 804761b..3e64742 100644
--- a/frontend/lib/api.ts
+++ b/frontend/lib/api.ts
@@ -236,6 +236,16 @@ export function listAudioEntries(): Promise<AudioEntryOut[]> {
   return apiFetch<AudioEntryOut[]>('/api/v1/audio');
 }
 
+export function renameAudioEntry(
+  id: string,
+  title: string,
+): Promise<AudioEntryOut> {
+  return apiFetch<AudioEntryOut>(`/api/v1/audio/${encodeURIComponent(id)}`, {
+    method: 'PATCH',
+    body: JSON.stringify({ title }),
+  });
+}
+
 export function deleteAudioEntry(id: string): Promise<void> {
   return apiFetchVoid(`/api/v1/audio/${encodeURIComponent(id)}`, {
     method: 'DELETE',

From 583a0a030eb6789334963dd85301572d9ebbaba0 Mon Sep 17 00:00:00 2001
From: krasi <krasimir.xristov@gmail.com>
Date: Sat, 16 May 2026 19:30:53 +0200
Subject: [PATCH 2/2] fix: address CodeRabbit review findings

- async temp I/O: wrap store/retrieve/delete_temp_audio in asyncio.to_thread
- extract rename_audio_entry to rename.py (service.py was ~460 lines)
- router: UUID token type + rate limit on /temp/{token} endpoint
- tool: await async temp audio service calls
- AudioCard: show optimisticTitle in delete dialog (was stale entry.title)
- useAudio: user-scoped cache keys to prevent cross-user cache leak
- api: Zod validation on renameAudioEntry inputs

skip: mask_pii intentionally removed (causes TTS to say 'person')
---
 backend/app/features/audio/rename.py          | 74 +++++++++++++++++
 backend/app/features/audio/router.py          |  8 +-
 backend/app/features/audio/service.py         | 82 ++++---------------
 backend/app/features/audio/tool.py            | 63 +++++++++-----
 .../features/audio/components/AudioCard.tsx   |  2 +-
 frontend/features/audio/hooks/useAudio.ts     | 22 ++++-
 frontend/lib/api.ts                           | 13 ++-
 7 files changed, 171 insertions(+), 93 deletions(-)
 create mode 100644 backend/app/features/audio/rename.py

diff --git a/backend/app/features/audio/rename.py b/backend/app/features/audio/rename.py
new file mode 100644
index 0000000..e2b018b
--- /dev/null
+++ b/backend/app/features/audio/rename.py
@@ -0,0 +1,74 @@
+"""rename_audio_entry — extracted from service.py to stay under the 300-line limit."""
+
+import logging
+from uuid import UUID
+
+from app.core.config import settings
+from app.db.supabase import get_supabase_client
+from app.features.audio.schemas import AudioEntryOut
+from app.features.audio.service import (
+    _SELECT_COLS,
+    _SIGNED_URL_TTL,
+    _row_to_entry,
+)
+
+logger = logging.getLogger(__name__)
+
+
+async def rename_audio_entry(
+    entry_id: UUID, user_id: str, title: str
+) -> AudioEntryOut | None:
+    """Rename an audio entry's title.
+
+    Returns the updated AudioEntryOut, or None if the entry was not found.
+    Raises RuntimeError on DB or Storage failure.
+    """
+    client = await get_supabase_client()
+
+    # UPDATE — postgrest-py's FilterRequestBuilder has no .select() after .eq(),
+    # so we do the update and re-fetch in a separate query.
+    try:
+        await (
+            client.table("audio_entries")
+            .update({"title": title})
+            .eq("id", str(entry_id))
+            .eq("user_id", user_id)
+            .execute()
+        )
+    except Exception:
+        logger.exception("[rename_audio_entry] UPDATE failed id=%s", entry_id)
+        raise RuntimeError("Failed to rename audio entry.") from None
+
+    # Re-fetch the updated row (also confirms the entry belongs to this user).
+    try:
+        fetch = (
+            await client.table("audio_entries")
+            .select(_SELECT_COLS)
+            .eq("id", str(entry_id))
+            .eq("user_id", user_id)
+            .execute()
+        )
+    except Exception:
+        logger.exception(
+            "[rename_audio_entry] SELECT after UPDATE failed id=%s", entry_id
+        )
+        raise RuntimeError("Failed to fetch renamed audio entry.") from None
+
+    if not fetch.data:
+        return None
+
+    row = fetch.data[0]
+    filename: str = row["filename"]
+    safe_log = filename.split("/")[-1] if "/" in filename else filename
+
+    try:
+        result = await client.storage.from_(settings.audio_bucket).create_signed_url(
+            filename, _SIGNED_URL_TTL
+        )
+        signed_url: str = result["signedURL"]
+    except Exception:
+        logger.exception("[rename_audio_entry] signed URL failed uuid=%s", safe_log)
+        raise RuntimeError("Title updated but failed to generate signed URL.") from None
+
+    logger.info("[rename_audio_entry] renamed uuid=%s", safe_log)
+    return _row_to_entry(row, signed_url)
diff --git a/backend/app/features/audio/router.py b/backend/app/features/audio/router.py
index 0f51151..757bfc8 100644
--- a/backend/app/features/audio/router.py
+++ b/backend/app/features/audio/router.py
@@ -10,6 +10,7 @@
 from app.core.security import get_current_user
 from app.features.auth.schemas import UserSchema
 from app.features.audio import service
+from app.features.audio.rename import rename_audio_entry as _rename_audio_entry
 from app.features.audio.schemas import AudioEntryOut, AudioEntryPatch
 
 logger = logging.getLogger(__name__)
@@ -17,14 +18,15 @@
 
 
 @router.get("/temp/{token}")
-async def serve_temp_audio(token: str) -> Response:
+@limiter.limit("30/minute")
+async def serve_temp_audio(token: UUID, request: Request) -> Response:
     """Serve a temporarily stored mp3 before the user confirms saving.
 
     No JWT auth — the UUID token itself acts as a short-lived capability URL
     (unguessable, 1-hour TTL).  The browser <audio> element cannot send custom
     headers, so cookie / bearer auth is not applicable here.
     """
-    audio_bytes = service.retrieve_temp_audio(token)
+    audio_bytes = await service.retrieve_temp_audio(str(token))
     if audio_bytes is None:
         raise HTTPException(
             status_code=status.HTTP_404_NOT_FOUND,
@@ -68,7 +70,7 @@ async def rename_audio_entry_endpoint(
     current_user: UserSchema = Depends(get_current_user),
 ) -> AudioEntryOut:
     """Rename a saved audio entry's title."""
-    updated = await service.rename_audio_entry(entry_id, current_user.id, body.title)
+    updated = await _rename_audio_entry(entry_id, current_user.id, body.title)
     if updated is None:
         raise HTTPException(
             status_code=status.HTTP_404_NOT_FOUND,
diff --git a/backend/app/features/audio/service.py b/backend/app/features/audio/service.py
index 4df2d78..6d1b0df 100644
--- a/backend/app/features/audio/service.py
+++ b/backend/app/features/audio/service.py
@@ -64,29 +64,33 @@
 _TEMP_TTL = 3600  # 1 hour
 
 
-def store_temp_audio(audio_bytes: bytes) -> str:
+async def store_temp_audio(audio_bytes: bytes) -> str:
     """Write mp3 bytes to a temp file and return a UUID token."""
     token = str(uuid4())
     path = _TEMP_DIR / f"{_TEMP_PREFIX}{token}.mp3"
-    path.write_bytes(audio_bytes)
+    await asyncio.to_thread(path.write_bytes, audio_bytes)
     return token
 
 
-def retrieve_temp_audio(token: str) -> bytes | None:
+async def retrieve_temp_audio(token: str) -> bytes | None:
     """Return mp3 bytes for the given token, or None if missing / expired."""
     path = _TEMP_DIR / f"{_TEMP_PREFIX}{token}.mp3"
-    if not path.exists():
-        return None
-    if time.time() - path.stat().st_mtime > _TEMP_TTL:
-        path.unlink(missing_ok=True)
-        return None
-    return path.read_bytes()
 
+    def _read() -> bytes | None:
+        if not path.exists():
+            return None
+        if time.time() - path.stat().st_mtime > _TEMP_TTL:
+            path.unlink(missing_ok=True)
+            return None
+        return path.read_bytes()
+
+    return await asyncio.to_thread(_read)
 
-def delete_temp_audio(token: str) -> None:
+
+async def delete_temp_audio(token: str) -> None:
     """Delete the temp file after a successful Storage upload."""
     path = _TEMP_DIR / f"{_TEMP_PREFIX}{token}.mp3"
-    path.unlink(missing_ok=True)
+    await asyncio.to_thread(lambda: path.unlink(missing_ok=True))
 
 
 async def generate_tts(text: str, user_id: str) -> tuple[bytes, str]:
@@ -391,60 +395,4 @@ async def delete_audio_entry(entry_id: UUID, user_id: str) -> bool:
     return True
 
 
-async def rename_audio_entry(
-    entry_id: UUID, user_id: str, title: str
-) -> AudioEntryOut | None:
-    """Rename an audio entry's title.
-
-    Returns the updated AudioEntryOut, or None if the entry was not found.
-    Raises RuntimeError on DB or Storage failure.
-    """
-    client = await get_supabase_client()
-
-    # UPDATE — postgrest-py's FilterRequestBuilder has no .select() after .eq(),
-    # so we do the update and re-fetch in a separate query.
-    try:
-        await (
-            client.table("audio_entries")
-            .update({"title": title})
-            .eq("id", str(entry_id))
-            .eq("user_id", user_id)
-            .execute()
-        )
-    except Exception:
-        logger.exception("[rename_audio_entry] UPDATE failed id=%s", entry_id)
-        raise RuntimeError("Failed to rename audio entry.") from None
 
-    # Re-fetch the updated row (also confirms the entry belongs to this user).
-    try:
-        fetch = (
-            await client.table("audio_entries")
-            .select(_SELECT_COLS)
-            .eq("id", str(entry_id))
-            .eq("user_id", user_id)
-            .execute()
-        )
-    except Exception:
-        logger.exception(
-            "[rename_audio_entry] SELECT after UPDATE failed id=%s", entry_id
-        )
-        raise RuntimeError("Failed to fetch renamed audio entry.") from None
-
-    if not fetch.data:
-        return None
-
-    row = fetch.data[0]
-    filename: str = row["filename"]
-    safe_log = filename.split("/")[-1] if "/" in filename else filename
-
-    try:
-        result = await client.storage.from_(settings.audio_bucket).create_signed_url(
-            filename, _SIGNED_URL_TTL
-        )
-        signed_url: str = result["signedURL"]
-    except Exception:
-        logger.exception("[rename_audio_entry] signed URL failed uuid=%s", safe_log)
-        raise RuntimeError("Title updated but failed to generate signed URL.") from None
-
-    logger.info("[rename_audio_entry] renamed uuid=%s", safe_log)
-    return _row_to_entry(row, signed_url)
diff --git a/backend/app/features/audio/tool.py b/backend/app/features/audio/tool.py
index 8158ce8..ecdb418 100644
--- a/backend/app/features/audio/tool.py
+++ b/backend/app/features/audio/tool.py
@@ -1,13 +1,14 @@
 """Generate-TTS and Save-Audio LangGraph nodes (Phase 10C + 10D).
 
-generate_tts_node  — synthesizes text to speech via OpenRouter TTS, uploads the
-                     mp3 to Supabase Storage, stores the result in
-                     ``state["pending_audio"]`` for the cross-turn save
-                     confirmation, and returns a ToolMessage with a markdown link.
+generate_tts_node  — synthesizes text to speech via OpenRouter TTS, writes the
+                     mp3 bytes to a system temp file (NOT in Supabase Storage),
+                     stores a temp token in ``state["pending_audio"]``
+                     for the cross-turn save confirmation, and returns a
+                     ToolMessage with a playable /audio/temp/{token} URL.
 
-save_audio_entry_node — reads ``state["pending_audio"]``, persists the audio
-                        entry to ``audio_entries``, and returns a ToolMessage
-                        confirming the save.
+save_audio_entry_node — reads ``state["pending_audio"]``, retrieves the bytes
+                        from the temp store, uploads to Supabase Storage, saves
+                        the DB entry, and returns a ToolMessage confirming the save.
 
 Neither node is called via LangChain tool execution — both are invoked directly
 by the orchestrator graph (same pattern as save_transcript_node).
@@ -27,13 +28,15 @@
 
 
 async def generate_tts_node(state: AxonState) -> dict:
-    """LangGraph node: synthesize text to speech and upload to Storage.
+    """LangGraph node: synthesize text to speech and store bytes in temp store.
 
     Reads the ``text`` argument from the supervisor's ``generate_tts`` tool_call,
     calls the audio service, and returns:
-    - A ToolMessage closing the open tool_call (required for valid message history).
-    - ``pending_audio`` state update with JSON ``{filename, signed_url, text_preview}``
+    - A ToolMessage with a playable /api/v1/audio/temp/{token} URL.
+    - ``pending_audio`` state update with ``{filename, temp_token, text_preview}``
       for the Phase 10D save confirmation flow.
+
+    The mp3 is NOT uploaded to Supabase Storage until the user confirms saving.
     """
     # -- Resolve tool_call_id and text arg -----------------------------------
     tool_call_id: str | None = None
@@ -63,10 +66,10 @@ def _error(msg: str) -> dict:
 
     user_id: str = state["user_id"]
 
-    # -- Generate TTS + upload ------------------------------------------------
+    # -- Generate TTS + store in temp (no Storage upload yet) ---------------
     try:
         audio_bytes, filename = await audio_service.generate_tts(text, user_id)
-        signed_url = await audio_service.upload_audio(audio_bytes, filename)
+        temp_token = await audio_service.store_temp_audio(audio_bytes)
     except RuntimeError as exc:
         logger.error("[generate_tts_node] service error: %s", exc)
         return _error(f"Audio generation failed: {exc}")
@@ -79,17 +82,22 @@ def _error(msg: str) -> dict:
     # -- Build pending_audio payload for Phase 10D save ----------------------
     pending: dict[str, str] = {
         "filename": filename,
-        "signed_url": signed_url,
+        "temp_token": temp_token,
         "text_preview": text[:100],
     }
 
+    temp_url = f"/api/v1/audio/temp/{temp_token}"
     result_msg = (
         f"Audio generated successfully.\n\n"
-        f"[Play audio]({signed_url})\n\n"
+        f"[Play audio]({temp_url})\n\n"
         f"Would you like me to save this to your audio library?"
     )
 
-    logger.info("[generate_tts_node] audio ready filename=%s", filename)
+    logger.info(
+        "[generate_tts_node] audio ready token=%s filename=%s",
+        temp_token[:8],
+        filename,
+    )
 
     messages = []
     if tool_call_id:
@@ -140,8 +148,9 @@ def _error(msg: str) -> dict:
         )
 
     filename: str = pending_audio.get("filename", "")
-    if not filename:
-        return _error("Audio data is missing the filename — please try again.")
+    temp_token: str = pending_audio.get("temp_token", "")
+    if not filename or not temp_token:
+        return _error("Audio data is incomplete — please regenerate and try again.")
 
     # Fall back to text_preview if no title was provided.
     if not title:
@@ -149,8 +158,21 @@ def _error(msg: str) -> dict:
 
     user_id: str = state["user_id"]
 
-    # -- Persist ------------------------------------------------------------
+    # -- Retrieve bytes from temp store -------------------------------------
+    audio_bytes = await audio_service.retrieve_temp_audio(temp_token)
+    if audio_bytes is None:
+        logger.warning(
+            "[save_audio_entry_node] temp audio expired token=%s",
+            temp_token[:8],
+        )
+        return _error(
+            "The audio preview has expired (> 1 hour). "
+            "Please regenerate the audio and try saving again."
+        )
+
+    # -- Upload to Storage + persist DB entry (both or neither) ---------------
     try:
+        await audio_service.upload_audio(audio_bytes, filename)
         saved = await audio_service.save_audio_entry(
             user_id=user_id,
             filename=filename,
@@ -158,12 +180,15 @@ def _error(msg: str) -> dict:
             source_type="custom",
         )
     except RuntimeError as exc:
-        logger.error("[save_audio_entry_node] service error: %s", exc)
+        logger.error("[save_audio_entry_node] upload/save error: %s", exc)
         return _error(f"Failed to save audio: {exc}")
     except Exception:
         logger.exception("[save_audio_entry_node] unexpected error")
         return _error("An unexpected error occurred while saving the audio.")
 
+    # -- Clean up temp store -------------------------------------------------
+    await audio_service.delete_temp_audio(temp_token)
+
     result_msg = f"Audio saved to your library ✓ — **{saved.title}**"
     logger.info("[save_audio_entry_node] saved id=%s", saved.id)
 
diff --git a/frontend/features/audio/components/AudioCard.tsx b/frontend/features/audio/components/AudioCard.tsx
index 67dd46f..de1b5a9 100644
--- a/frontend/features/audio/components/AudioCard.tsx
+++ b/frontend/features/audio/components/AudioCard.tsx
@@ -157,7 +157,7 @@ const AudioCard = ({ entry }: AudioCardProps) => {
         <DialogPopup>
           <DialogTitle>Delete audio?</DialogTitle>
           <DialogDescription>
-            This will permanently delete &ldquo;{entry.title}&rdquo; from your
+            This will permanently delete &ldquo;{optimisticTitle}&rdquo; from your
             library and remove the file from storage. This action cannot be
             undone.
           </DialogDescription>
diff --git a/frontend/features/audio/hooks/useAudio.ts b/frontend/features/audio/hooks/useAudio.ts
index 31537af..6d6e596 100644
--- a/frontend/features/audio/hooks/useAudio.ts
+++ b/frontend/features/audio/hooks/useAudio.ts
@@ -1,19 +1,38 @@
 'use client';
 
+import { useEffect, useState } from 'react';
 import { useMutation, useQuery, useQueryClient } from '@tanstack/react-query';
 import {
   deleteAudioEntry,
   listAudioEntries,
   renameAudioEntry,
 } from '@/lib/api';
+import { createClient } from '@/lib/supabase/client';
 import type { AudioEntryOut } from '@/features/audio/types';
 
+// Broad prefix used for mutation invalidation — matches all user variants.
 export const AUDIO_ENTRIES_KEY = ['audio-entries'] as const;
 
+function audioEntriesKey(userId: string) {
+  return ['audio-entries', userId] as const;
+}
+
+function useUserId(): string {
+  const [userId, setUserId] = useState('');
+  useEffect(() => {
+    createClient()
+      .auth.getSession()
+      .then(({ data }) => setUserId(data.session?.user.id ?? ''));
+  }, []);
+  return userId;
+}
+
 export function useAudioEntries() {
+  const userId = useUserId();
   return useQuery<AudioEntryOut[]>({
-    queryKey: AUDIO_ENTRIES_KEY,
+    queryKey: audioEntriesKey(userId),
     queryFn: listAudioEntries,
+    enabled: !!userId,
     staleTime: 60_000,
   });
 }
@@ -37,3 +56,4 @@ export function useRenameAudio() {
     },
   });
 }
+
diff --git a/frontend/lib/api.ts b/frontend/lib/api.ts
index 3e64742..437a708 100644
--- a/frontend/lib/api.ts
+++ b/frontend/lib/api.ts
@@ -236,13 +236,22 @@ export function listAudioEntries(): Promise<AudioEntryOut[]> {
   return apiFetch<AudioEntryOut[]>('/api/v1/audio');
 }
 
+const _renameAudioSchema = z.object({
+  id: z.string().uuid(),
+  title: z.string().min(1).max(200).trim(),
+});
+
 export function renameAudioEntry(
   id: string,
   title: string,
 ): Promise<AudioEntryOut> {
-  return apiFetch<AudioEntryOut>(`/api/v1/audio/${encodeURIComponent(id)}`, {
+  const { id: validId, title: validTitle } = _renameAudioSchema.parse({
+    id,
+    title,
+  });
+  return apiFetch<AudioEntryOut>(`/api/v1/audio/${encodeURIComponent(validId)}`, {
     method: 'PATCH',
-    body: JSON.stringify({ title }),
+    body: JSON.stringify({ title: validTitle }),
   });
 }