diff --git a/README.md b/README.md
index f65d42b..4243ed7 100644
--- a/README.md
+++ b/README.md
@@ -22,10 +22,6 @@ candidate direct-download URLs from Civitai and Hugging Face.
   ComfyUI model folder.
 - Shows candidate file sizes and loads successfully downloaded files into their workflow nodes.
 - Searches the configured Quark Netdisk libraries directly for missing model filenames.
-- Audits existing model locations and reports only files whose expected official folder
-  can be determined clearly. The audit never moves files.
-- Leaves ambiguous checkpoint/diffusion model files in place instead of guessing their
-  architecture from names such as `flux`, `wan`, or `z_image`.
 - Ignores serialized workflow metadata, URLs, free-text fields, and generic `model`
   fields on non-loader custom nodes to prevent false missing-model reports.
 - Does not move, delete, or overwrite model files.
@@ -53,7 +49,7 @@ No extra Python packages are required.
 3. Review local replacement suggestions.
 4. Click **一键加载模型** to load safe high-confidence local matches into their nodes, or apply an
    individual suggestion.
-5. For missing models, click **查找下载来源**. You can download a Civitai or Hugging
+5. For missing models, click **下载缺失模型**. You can download a Civitai or Hugging
    Face candidate or a public match from the two configured Quark Netdisk model libraries
    directly into its matching configured model folder.
 
@@ -61,9 +57,10 @@ The download results are search candidates. Verify the model page, license, base
 and filename before downloading. Some Civitai files require authentication, so their
 direct URL may redirect to a sign-in page.
 
-Only sources whose filename exactly matches the missing workflow filename and whose
-download endpoint is currently usable are shown. Similar filenames and restricted Quark
-files are excluded.
+Only sources whose filename exactly matches the missing workflow filename are shown.
+Matching Quark files are shown as download buttons instead of fallback search links.
+The download endpoint rejects unknown model categories rather than placing files in an
+incorrect folder.
 
 Downloads are accepted only from approved Civitai and Hugging Face HTTPS hosts. Existing
 files are never overwritten. Completed downloads clear ComfyUI's model filename cache so
@@ -91,7 +88,7 @@ python -m unittest discover -s tests -v
 ## Privacy
 
 Workflow model filenames are sent to Civitai, Hugging Face, and the two configured public
-Quark shares only when you explicitly click **查找下载来源**. Regular scans remain local.
+Quark shares only when you explicitly click **下载缺失模型**. Regular scans remain local.
 
 ## License
 
diff --git a/find_models.py b/find_models.py
index 4386319..9266e2c 100644
--- a/find_models.py
+++ b/find_models.py
@@ -55,19 +55,6 @@ DOWNLOAD_CATEGORY_ALIASES = {
     "upscale_models": ("upscale_models",),
     "embeddings": ("embeddings",),
 }
-OFFICIAL_MODEL_FOLDERS = {
-    "checkpoints": "checkpoints",
-    "loras": "loras",
-    "vae": "vae",
-    "controlnet": "controlnet",
-    "clip_vision": "clip_vision",
-    "text_encoders": "text_encoders",
-    "diffusion_models": "diffusion_models",
-    "upscale_models": "upscale_models",
-    "embeddings": "embeddings",
-}
-
-
 def _safe_query(name: str) -> str:
     return re.sub(r"[_\-.]+", " ", basename(name).rsplit(".", 1)[0]).strip()
 
@@ -118,7 +105,10 @@ def _is_model_payload(path: Path) -> bool:
 
 
 def _target_directory(category: str) -> Path:
-    for candidate in DOWNLOAD_CATEGORY_ALIASES.get(category, (category,)):
+    aliases = DOWNLOAD_CATEGORY_ALIASES.get(category)
+    if not aliases:
+        raise web.HTTPBadRequest(text=f"无法确定模型对应的 ComfyUI 文件夹: {category}")
+    for candidate in aliases:
         try:
             paths = folder_paths.get_folder_paths(candidate)
         except Exception:
@@ -137,54 +127,6 @@ def _clear_filename_cache(category: str) -> None:
             cache.pop(candidate, None)
 
 
-def _comfy_model_root() -> Path:
-    base = getattr(folder_paths, "models_dir", None)
-    if not base:
-        checkpoint_paths = folder_paths.get_folder_paths("checkpoints")
-        if not checkpoint_paths:
-            raise web.HTTPBadRequest(text="无法确定 ComfyUI models 目录")
-        base = Path(checkpoint_paths[0]).parent
-    return Path(base).resolve()
-
-
-def _classify_existing_file(path: Path) -> str | None:
-    text = path.as_posix().lower()
-    name = path.name.lower()
-    if any(term in text for term in ("lora", "lycoris")):
-        return "loras"
-    if "controlnet" in text or "control_net" in text:
-        return "controlnet"
-    if "clip_vision" in text:
-        return "clip_vision"
-    if any(term in text for term in ("text_encoder", "/clip/", "/t5")):
-        return "text_encoders"
-    if "embedding" in text:
-        return "embeddings"
-    if any(term in text for term in ("upscale", "esrgan", "rife", "frame_interpolation")):
-        return "upscale_models"
-    if "/vae" in text or name.startswith("vae") or "_vae" in name:
-        return "vae"
-    if any(term in text for term in ("diffusion_model", "/unet/")):
-        return "diffusion_models"
-    return None
-
-
-def _audit_model_locations() -> list[dict[str, str]]:
-    root = _comfy_model_root()
-    issues = []
-    for source in root.rglob("*"):
-        if not source.is_file() or source.suffix.lower() not in MODEL_EXTENSIONS:
-            continue
-        category = _classify_existing_file(source)
-        if not category:
-            continue
-        target_dir = (root / OFFICIAL_MODEL_FOLDERS[category]).resolve()
-        if source.parent == target_dir or target_dir in source.parents:
-            continue
-        issues.append({"expected": category, "path": str(source)})
-    return issues
-
-
 async def _get_json(session: aiohttp.ClientSession, url: str) -> Any:
     try:
         async with session.get(url, headers={"User-Agent": "ComfyUI_FindModels/1.0"}) as response:
@@ -387,20 +329,10 @@ async def find_sources(request: web.Request) -> web.Response:
         *(_validate_web_candidate(session, candidate) for candidate in exact_web),
         return_exceptions=True,
     )
-    checked_quark = await asyncio.gather(
-        *(_validate_quark_candidate(session, candidate) for candidate in exact_quark),
-        return_exceptions=True,
-    )
     candidates = [
-        candidate for candidate in [*checked_web, *checked_quark] if isinstance(candidate, dict)
+        candidate for candidate in [*checked_web, *exact_quark] if isinstance(candidate, dict)
     ]
-    return web.json_response({"name": name, "candidates": candidates[:12], "quark_libraries": QUARK_MODEL_LIBRARIES})
-
-
-@PromptServer.instance.routes.get("/findmodels/audit")
-async def audit_models(request: web.Request) -> web.Response:
-    items = _audit_model_locations()
-    return web.json_response({"count": len(items), "items": items})
+    return web.json_response({"name": name, "candidates": candidates[:12]})
 
 
 async def _quark_download_url(session: aiohttp.ClientSession, payload: dict[str, Any]) -> str:
@@ -429,17 +361,7 @@ async def _quark_download_url(session: aiohttp.ClientSession, payload: dict[str,
         if _allowed_quark_download_url(url):
             return url
     detail = errors[0] if errors else "需要登录或文件超过公开下载大小限制"
-    raise web.HTTPBadGateway(text=f"夸克无法公开下载：{detail}")
-
-
-async def _validate_quark_candidate(
-    session: aiohttp.ClientSession, candidate: dict[str, Any]
-) -> dict[str, Any] | None:
-    try:
-        await _quark_download_url(session, candidate["quark"])
-        return candidate
-    except web.HTTPException:
-        return None
+    raise web.HTTPBadGateway(text=f"夸克拒绝直链下载：{detail}")
 
 
 @PromptServer.instance.routes.post("/findmodels/download")
diff --git a/model_finder.py b/model_finder.py
index 4604cee..4a16855 100644
--- a/model_finder.py
+++ b/model_finder.py
@@ -121,7 +121,7 @@ class Reference:
 
 
 def normalize_path(value: str) -> str:
-    return value.replace("\\", "/").strip()
+    return value.replace("\\", "/").strip().lstrip("/")
 
 
 def basename(value: str) -> str:
diff --git a/pyproject.toml b/pyproject.toml
index ea5229d..c70f61b 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "comfyui-findmodels"
-version = "1.7.0"
+version = "1.8.0"
 description = "Find and adapt missing models in the current ComfyUI workflow"
 readme = "README.md"
 requires-python = ">=3.10"
diff --git a/tests/test_download_helpers.py b/tests/test_download_helpers.py
index 135f9da..7130ad0 100644
--- a/tests/test_download_helpers.py
+++ b/tests/test_download_helpers.py
@@ -33,8 +33,6 @@ from ComfyUI_FindModels.find_models import (
     _safe_filename,
     _size_value,
     _target_directory,
-    _classify_existing_file,
-    _audit_model_locations,
 )
 
 
@@ -76,6 +74,10 @@ class DownloadHelperTests(unittest.TestCase):
     def test_uses_matching_comfyui_model_folder(self):
         self.assertEqual(_target_directory("loras"), (Path.cwd() / "models" / "loras").resolve())
 
+    def test_rejects_unknown_download_folder(self):
+        with self.assertRaises(Exception):
+            _target_directory("unknown")
+
     def test_supports_legacy_comfyui_folder_aliases(self):
         original = folder_paths.get_folder_paths
         folder_paths.get_folder_paths = lambda category: [] if category == "text_encoders" else [str(Path.cwd() / "models" / category)]
@@ -99,29 +101,5 @@ class DownloadHelperTests(unittest.TestCase):
             ],
         )
 
-    def test_classifies_files_for_official_folders(self):
-        self.assertEqual(_classify_existing_file(Path("misc/loras/hero.safetensors")), "loras")
-        self.assertEqual(_classify_existing_file(Path("misc/rife47.pth")), "upscale_models")
-        self.assertEqual(_classify_existing_file(Path("misc/vae_model.safetensors")), "vae")
-        self.assertIsNone(_classify_existing_file(Path("checkpoints/z_image_turbo_bf16.safetensors")))
-        self.assertIsNone(_classify_existing_file(Path("misc/unknown_model.safetensors")))
-
-    def test_audit_reports_only_clearly_misplaced_files(self):
-        root = Path(folder_paths.models_dir)
-        misplaced = root / "checkpoints" / "hero_lora.safetensors"
-        correct = root / "loras" / "people" / "correct.safetensors"
-        misplaced.parent.mkdir(parents=True, exist_ok=True)
-        correct.parent.mkdir(parents=True, exist_ok=True)
-        misplaced.write_bytes(b"x")
-        correct.write_bytes(b"x")
-        try:
-            issues = _audit_model_locations()
-            self.assertTrue(any(item["path"] == str(misplaced.resolve()) for item in issues))
-            self.assertFalse(any(item["path"] == str(correct.resolve()) for item in issues))
-        finally:
-            misplaced.unlink()
-            correct.unlink()
-
-
 if __name__ == "__main__":
     unittest.main()
diff --git a/tests/test_model_finder.py b/tests/test_model_finder.py
index 2e8609b..8297cff 100644
--- a/tests/test_model_finder.py
+++ b/tests/test_model_finder.py
@@ -27,6 +27,12 @@ class ModelFinderTests(unittest.TestCase):
     def test_normalized_stem_removes_common_precision_suffix(self):
         self.assertEqual(normalized_stem("Hero_v2_fp16.safetensors"), "hero")
 
+    def test_removes_leading_path_separators(self):
+        refs = extract_references(
+            {"nodes": [{"id": 6, "type": "VAELoader", "widgets": [{"name": "vae_name", "value": "\\/folder/model.safetensors"}]}]}
+        )
+        self.assertEqual(refs[0].name, "folder/model.safetensors")
+
     def test_extracts_widget_reference_with_location(self):
         refs = extract_references(
             {"nodes": [{"id": 7, "type": "LoraLoader", "widgets": [{"name": "lora_name", "value": "Hero.safetensors"}]}]}
diff --git a/web/find_models.js b/web/find_models.js
index dd47b12..1014965 100644
--- a/web/find_models.js
+++ b/web/find_models.js
@@ -47,12 +47,6 @@ async function post(path, body) {
   return response.json();
 }
 
-async function get(path) {
-  const response = await api.fetchApi(path);
-  if (!response.ok) throw new Error(await response.text());
-  return response.json();
-}
-
 function findTargetWidget(node, model) {
   const widgets = node?.widgets || [];
   return widgets.find((item) => item.name === model.widget)
@@ -112,7 +106,6 @@ function ensurePanel() {
     <div class="fm-actions">
       <button data-action="scan">扫描当前工作流</button>
       <button data-action="adapt">一键加载模型</button>
-      <button data-action="audit">检查模型位置</button>
     </div>
     <div class="fm-summary">尚未扫描</div><div class="fm-list"></div>`;
   document.body.appendChild(panel);
@@ -133,31 +126,11 @@ function ensurePanel() {
     button.disabled = false;
     if (count) window.setTimeout(() => scan(false), 100);
   };
-  panel.querySelector('[data-action="audit"]').onclick = async () => {
-    const button = panel.querySelector('[data-action="audit"]');
-    button.disabled = true;
-    try {
-      const audit = await get("/findmodels/audit");
-      if (!audit.count) {
-        panel.querySelector(".fm-summary").textContent = "未发现能够明确判断为放错位置的模型。";
-        return;
-      }
-      const preview = audit.items.slice(0, 50).map((item) =>
-        `${item.expected}: ${item.path}`,
-      ).join("\n");
-      window.alert(`发现 ${audit.count} 个能够明确判断的位置问题。\n本功能只检查，不移动文件。\n\n${preview}`);
-      panel.querySelector(".fm-summary").textContent = `发现 ${audit.count} 个明确的位置问题，未移动任何文件。`;
-    } catch (error) {
-      panel.querySelector(".fm-summary").textContent = `检查失败：${error.message}`;
-    } finally {
-      button.disabled = false;
-    }
-  };
   panel.addEventListener("click", async (event) => {
     const sourceButton = event.target.closest("[data-source]");
     if (sourceButton) {
       sourceButton.disabled = true;
-      sourceButton.textContent = "查询中…";
+      sourceButton.textContent = "正在获取下载项…";
       const model = lastResult?.models?.find((item) => item.name === sourceButton.dataset.source);
       const target = sourceButton.closest(".fm-item").querySelector(".fm-sources");
       try {
@@ -166,15 +139,12 @@ function ensurePanel() {
           category: model.category,
         });
         const links = data.candidates.map((item) => sourceHtml(item, model)).join("");
-        const quarkFallback = data.quark_libraries.map((item) =>
-          `<a href="${escapeHtml(item.url)}" target="_blank" rel="noopener noreferrer">${escapeHtml(item.name)}备用查找</a>`,
-        ).join("");
-        target.innerHTML = `${links || "<span>未找到文件名完全一致且可直接下载的模型</span>"}${quarkFallback}`;
+        target.innerHTML = links || "<span>未找到文件名完全一致的可下载模型</span>";
       } catch (error) {
-        target.textContent = `查询失败：${error.message}`;
+        target.textContent = `获取下载项失败：${error.message}`;
       } finally {
         sourceButton.disabled = false;
-        sourceButton.textContent = "查找下载来源";
+        sourceButton.textContent = "下载缺失模型";
       }
       return;
     }
@@ -219,7 +189,7 @@ function render(result, quiet) {
     <article class="fm-item fm-${escapeHtml(model.status)}">
       <div><strong>${escapeHtml(model.name)}</strong><span>${escapeHtml(model.category)} · ${escapeHtml(model.status)}</span></div>
       ${model.match ? `<div class="fm-match">本地候选：${escapeHtml(model.match.name)} (${Math.round(model.match.confidence * 100)}%) ${model.match.auto_apply ? `<button data-apply="${escapeHtml(model.node_id)}:${escapeHtml(model.widget)}">加载</button>` : ""}</div>` : ""}
-      ${model.status === "missing" ? `<button data-source="${escapeHtml(model.name)}">查找下载来源</button><div class="fm-sources"></div>` : ""}
+      ${model.status === "missing" ? `<button data-source="${escapeHtml(model.name)}">下载缺失模型</button><div class="fm-sources"></div>` : ""}
     </article>`).join("") || "<p>当前工作流中未识别到模型引用。</p>";
   panel.querySelectorAll("[data-apply]").forEach((button) => {
     button.onclick = async () => {