zen-gitsync 2.17.5 → 2.17.6

package/src/utils/aiEndpoint.js

@@ -0,0 +1,231 @@
+// Copyright 2026 xz333221
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+//
+// ─────────────────────────────────────────────────────────────
+// AI 请求的地址 + 请求头统一构造层
+//
+// 为什么要单独抽一层（以前这些逻辑散在 7 个调用点，只改一处等于没改）：
+//
+// 1) 客户端身份 + 会话 ID。OpenCode 的网关（opencode.ai/zen/**，含 $10/月的 Go 订阅）
+//    要求客户端「用自己的名字当 User-Agent」（不能是 node-fetch/undici 这类通用库名），
+//    并且每次对话在 `x-opencode-session` 里带一个稳定会话 ID，用于路由与提示缓存。
+//    缺了会被网关直接拒掉：
+//      Request is missing x-opencode-session and cannot be routed efficiently.
+//      Please see https://opencode.ai/docs/go/#where-can-i-use-it
+//    用 node-fetch 默认 UA 也会被当成"通用 HTTP 库"而命中滥用风控。
+//
+// 2) 协议族选择。同一个 baseURL 后面挂着三种协议，path 与鉴权方式都不同：
+//      chat      → POST /chat/completions  (Authorization: Bearer)
+//      messages  → POST /messages          (x-api-key + anthropic-version)
+//      responses → POST /responses         (Authorization: Bearer)
+//    本仓库的 AI 调用全部只实现了 OpenAI 兼容的 chat 形态，所以碰到后两种协议族
+//    要**明确报错**（提示换模型），而不是把 OpenAI body 打到 Anthropic 路由上等一个
+//    看不懂的 4xx。
+//
+// 参考实现：ai-model-form 包的 server/middleware.js（同一套判断，那边只做连通性测试）。
+// ─────────────────────────────────────────────────────────────
+
+import { randomUUID } from 'node:crypto'
+import { readFileSync } from 'node:fs'
+
+/** 本包版本，用于 User-Agent（从模块自身位置往上找 package.json，不受 cwd 影响） */
+const PKG_VERSION = (() => {
+  try {
+    const pkg = JSON.parse(readFileSync(new URL('../../package.json', import.meta.url), 'utf8'))
+    return pkg.version || '0.0.0'
+  } catch {
+    return '0.0.0'
+  }
+})()
+
+/** 网关/风控认的客户端标识：得是"自己的名字"，不能是通用 HTTP 库 */
+export const AI_USER_AGENT = `zen-gitsync/${PKG_VERSION}`
+
+/** OpenCode 网关的 Go 订阅入口（ai-model-form 里叫 OPENCODE_GO_URL） */
+export const OPENCODE_GO_URL = 'https://opencode.ai/zen/go/v1'
+
+/**
+ * OpenCode Go 上走 Anthropic /messages 协议的模型（鉴权是 x-api-key，不是 Bearer）。
+ * 只列**非 chat** 的模型：没列到的一律按 chat 处理，这样网关以后新增 chat 模型
+ * 不需要同步改这里（新增非 chat 模型时才会退化回"上游报错"）。
+ */
+const OPENCODE_GO_MESSAGES_MODELS = new Set([
+  'minimax-m3', 'minimax-m2.7', 'minimax-m2.5',
+  'qwen3.8-max', 'qwen3.8-flash', 'qwen3.7-max', 'qwen3.7-plus',
+  'qwen3.6-plus', 'qwen3.5-plus', 'union-alpha',
+])
+
+/** OpenCode Go 上走 OpenAI Responses 协议（/responses）的模型 */
+const OPENCODE_GO_RESPONSES_MODELS = new Set([
+  'grok-4.6', 'grok-4.5', 'gpt-5.6-luna',
+  'muse-spark-1.3-contributor', 'muse-spark-1.2-contributor',
+])
+
+/**
+ * 归一化 baseURL：去掉尾部斜杠（兼容用户填 .../v1/ 的写法）
+ * @param {string} baseURL
+ * @returns {string}
+ */
+export function normalizeBaseURL(baseURL) {
+  return String(baseURL || '').trim().replace(/\/+$/, '')
+}
+
+/**
+ * 是否为 OpenCode 网关（opencode.ai 及其子域）。
+ * 只有这类端点才需要客户端身份 + 会话 ID；其它 provider 不必多带这几个头。
+ * @param {string} baseURL
+ * @returns {boolean}
+ */
+export function isOpenCodeGateway(baseURL) {
+  try {
+    const host = new URL(normalizeBaseURL(baseURL)).hostname
+    return /(^|\.)opencode\.ai$/i.test(host)
+  } catch {
+    return false
+  }
+}
+
+/**
+ * 判断模型在 OpenCode Go 上走哪种协议族。
+ * @param {string} modelId
+ * @returns {'chat'|'messages'|'responses'}
+ */
+export function openCodeGoFamily(modelId) {
+  const id = String(modelId || '').trim().toLowerCase()
+  if (OPENCODE_GO_MESSAGES_MODELS.has(id)) return 'messages'
+  if (OPENCODE_GO_RESPONSES_MODELS.has(id)) return 'responses'
+  return 'chat'
+}
+
+let _processSessionId = null
+
+/**
+ * 进程级稳定会话 ID（懒创建）。
+ * 一次性调用（生成提交信息 / 分支名 / 代码分析…）没有"对话"概念，用同一个值把
+ * 同一进程里的这类请求归到一条会话上——重复点"重新生成"还能吃到网关的提示缓存。
+ * @returns {string}
+ */
+export function processAiSessionId() {
+  if (!_processSessionId) _processSessionId = randomUUID()
+  return _processSessionId
+}
+
+/**
+ * 请求的公共头：客户端身份（+ OpenCode 网关的会话 ID）。
+ * GET /models、连通性探测这类非 chat 请求也应该带上，否则同样会被风控挡。
+ *
+ * @param {object} opts
+ * @param {string} opts.baseURL
+ * @param {string} [opts.apiKey]
+ * @param {string} [opts.sessionId] - 对话级 ID；不传则用进程级兜底
+ * @returns {Record<string,string>}
+ */
+export function buildAiBaseHeaders({ baseURL, apiKey, sessionId } = {}) {
+  const headers = { 'User-Agent': AI_USER_AGENT }
+  if (isOpenCodeGateway(baseURL)) {
+    headers['x-opencode-session'] = sessionId || processAiSessionId()
+  }
+  if (apiKey) headers['Authorization'] = `Bearer ${apiKey}`
+  return headers
+}
+
+/**
+ * 构造「OpenAI 兼容 chat 请求」的 URL 与完整请求头。
+ *
+ * @param {object} opts
+ * @param {string} opts.baseURL
+ * @param {string} [opts.model]     - 用于识别 OpenCode Go 的协议族
+ * @param {string} [opts.apiKey]
+ * @param {string} [opts.sessionId] - 对话级 ID（多轮对话请传，一次性调用可省略）
+ * @param {Record<string,string>} [opts.extraHeaders]
+ * @returns {{ url: string, headers: Record<string,string> }}
+ * @throws {Error} baseURL 缺失，或所选模型在 OpenCode 网关上不是 chat/completions 协议
+ */
+export function buildAiChatRequest({ baseURL, model, apiKey, sessionId, extraHeaders } = {}) {
+  const base = normalizeBaseURL(baseURL)
+  if (!base) throw new Error('AI 模型未配置接口地址（baseURL）')
+
+  if (isOpenCodeGateway(base)) {
+    const family = openCodeGoFamily(model)
+    if (family !== 'chat') {
+      const proto = family === 'messages'
+        ? 'Anthropic 的 /messages（鉴权走 x-api-key）'
+        : 'OpenAI 的 /responses'
+      throw new Error(
+        `模型「${model}」在 OpenCode 网关上走 ${proto} 协议，本工具的 AI 调用只实现了 ` +
+        'OpenAI 兼容的 /chat/completions。请在设置里换一个走 chat/completions 的模型' +
+        '（如 deepseek-v4.1-flash、glm-5.3、kimi-k3）。'
+      )
+    }
+  }
+
+  return {
+    url: `${base}/chat/completions`,
+    headers: {
+      'Content-Type': 'application/json',
+      ...buildAiBaseHeaders({ baseURL: base, apiKey, sessionId }),
+      ...extraHeaders,
+    },
+  }
+}
+
+/**
+ * 把上游返回的错误体整理成一句人能读的话。
+ *
+ * 各家 provider 的错误体形态不一（OpenAI 兼容是 {error:{message}}，Anthropic 是
+ * {error:{message}} / {message}，还有些直接返回纯文本或 HTML），以前直接把整个响应体
+ * 丢进 Error 里，界面上就是一大坨 JSON。这里统一抽 message，抽不到才退回原文（截断）。
+ *
+ * @param {string} bodyText - 响应体原文
+ * @param {number} [status]
+ * @returns {string}
+ */
+export function describeAiHttpError(bodyText, status) {
+  const raw = String(bodyText || '').trim()
+  let message = ''
+  try {
+    const data = JSON.parse(raw)
+    message = data?.error?.message
+      || (typeof data?.error === 'string' ? data.error : '')
+      || data?.message
+      || ''
+  } catch {
+    message = ''
+  }
+  if (message) return message
+  const fallback = raw.length > 300 ? `${raw.slice(0, 300)}…` : raw
+  return fallback || `HTTP ${status}`
+}
+
+/**
+ * 构造非 chat 请求（GET /models 等）的 URL 与请求头。
+ * @param {object} opts
+ * @param {string} opts.baseURL
+ * @param {string} opts.path      - 以 / 开头的路径，如 '/models'
+ * @param {string} [opts.apiKey]
+ * @param {string} [opts.sessionId]
+ * @param {Record<string,string>} [opts.extraHeaders]
+ * @returns {{ url: string, headers: Record<string,string> }}
+ */
+export function buildAiApiRequest({ baseURL, path, apiKey, sessionId, extraHeaders } = {}) {
+  const base = normalizeBaseURL(baseURL)
+  if (!base) throw new Error('AI 模型未配置接口地址（baseURL）')
+  return {
+    url: `${base}${path}`,
+    headers: {
+      ...buildAiBaseHeaders({ baseURL: base, apiKey, sessionId }),
+      ...extraHeaders,
+    },
+  }
+}

package/src/utils/aiEndpoint.test.js

@@ -0,0 +1,175 @@
+// Copyright 2026 xz333221
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+//
+// src/utils/aiEndpoint.js 单元测试(node:test 内置)。
+//
+// 这一层是为了修「生成提交信息报 Request is missing x-opencode-session」而抽出来的，
+// 所以用测试把三件事钉死：
+//   1) 只对 OpenCode 网关加 x-opencode-session，别的 provider 不加（避免污染第三方请求）
+//   2) 请求头里必须有自己的客户端名字（User-Agent 不能是 node-fetch/undici）
+//   3) OpenCode 网关上非 chat 协议族的模型要**明确报错**，不能把 OpenAI body 打错路由
+import { test } from 'node:test'
+import assert from 'node:assert/strict'
+import {
+  AI_USER_AGENT,
+  OPENCODE_GO_URL,
+  buildAiApiRequest,
+  buildAiChatRequest,
+  describeAiHttpError,
+  isOpenCodeGateway,
+  normalizeBaseURL,
+  openCodeGoFamily,
+  processAiSessionId,
+} from './aiEndpoint.js'
+
+const GO = 'https://opencode.ai/zen/go/v1'
+
+// ========== normalizeBaseURL ==========
+
+test('normalizeBaseURL: 去掉尾部斜杠并 trim', () => {
+  assert.equal(normalizeBaseURL('https://api.deepseek.com/v1/'), 'https://api.deepseek.com/v1')
+  assert.equal(normalizeBaseURL('  https://api.deepseek.com/v1//  '), 'https://api.deepseek.com/v1')
+  assert.equal(normalizeBaseURL(''), '')
+  assert.equal(normalizeBaseURL(undefined), '')
+})
+
+// ========== isOpenCodeGateway ==========
+
+test('isOpenCodeGateway: 识别 opencode.ai 及其子域', () => {
+  assert.equal(isOpenCodeGateway(GO), true)
+  assert.equal(isOpenCodeGateway('https://opencode.ai/zen/go/v1/'), true)
+  assert.equal(isOpenCodeGateway('https://api.opencode.ai/v1'), true)
+  assert.equal(isOpenCodeGateway('https://api.deepseek.com/v1'), false)
+  // 不能被"看起来像"的域名骗到（endsWith 判断容易踩）
+  assert.equal(isOpenCodeGateway('https://evil-opencode.ai.attacker.com/v1'), false)
+  assert.equal(isOpenCodeGateway('not-a-url'), false)
+  assert.equal(isOpenCodeGateway(''), false)
+})
+
+// ========== openCodeGoFamily ==========
+
+test('openCodeGoFamily: 三种协议族与默认 chat', () => {
+  assert.equal(openCodeGoFamily('deepseek-v4.1-flash'), 'chat')
+  assert.equal(openCodeGoFamily('glm-5.3'), 'chat')
+  assert.equal(openCodeGoFamily('minimax-m3'), 'messages')
+  assert.equal(openCodeGoFamily('qwen3.8-max'), 'messages')
+  assert.equal(openCodeGoFamily('grok-4.6'), 'responses')
+  assert.equal(openCodeGoFamily('gpt-5.6-luna'), 'responses')
+  // 大小写/空格容错
+  assert.equal(openCodeGoFamily(' MiniMax-M3 '), 'messages')
+  // 未知/新增模型一律按 chat：网关以后加 chat 模型不需要同步改代码
+  assert.equal(openCodeGoFamily('brand-new-model'), 'chat')
+  assert.equal(openCodeGoFamily(''), 'chat')
+})
+
+// ========== buildAiChatRequest ==========
+
+test('buildAiChatRequest: 普通 provider 拼 url + Bearer，不带 opencode 会话头', () => {
+  const { url, headers } = buildAiChatRequest({
+    baseURL: 'https://api.deepseek.com/v1/',
+    model: 'deepseek-chat',
+    apiKey: 'sk-abc',
+  })
+  assert.equal(url, 'https://api.deepseek.com/v1/chat/completions')
+  assert.equal(headers['Content-Type'], 'application/json')
+  assert.equal(headers['Authorization'], 'Bearer sk-abc')
+  assert.equal(headers['User-Agent'], AI_USER_AGENT)
+  assert.ok(!('x-opencode-session' in headers), '非 OpenCode 网关不应带 x-opencode-session')
+})
+
+test('buildAiChatRequest: OpenCode Go + chat 模型带上客户端身份与会话 ID', () => {
+  const { url, headers } = buildAiChatRequest({
+    baseURL: GO,
+    model: 'deepseek-v4.1-flash',
+    apiKey: 'sk-go',
+  })
+  assert.equal(url, `${OPENCODE_GO_URL}/chat/completions`)
+  assert.equal(headers['Authorization'], 'Bearer sk-go')
+  assert.ok(headers['x-opencode-session'], '缺少 x-opencode-session 会被网关拒绝')
+  // 网关要求"用自己的名字"当 UA，通用库名（node-fetch/undici）会命中风控
+  assert.match(headers['User-Agent'], /^zen-gitsync\/\d+\.\d+\.\d+$/)
+})
+
+test('buildAiChatRequest: 省略 sessionId 时同进程内保持稳定（吃提示缓存）', () => {
+  const a = buildAiChatRequest({ baseURL: GO, model: 'deepseek-v4.1-flash', apiKey: 'k' })
+  const b = buildAiChatRequest({ baseURL: GO, model: 'deepseek-v4.1-flash', apiKey: 'k' })
+  assert.equal(a.headers['x-opencode-session'], b.headers['x-opencode-session'])
+  assert.equal(a.headers['x-opencode-session'], processAiSessionId())
+})
+
+test('buildAiChatRequest: 显式 sessionId 覆盖进程级兜底（对话级复用）', () => {
+  const { headers } = buildAiChatRequest({
+    baseURL: GO,
+    model: 'deepseek-v4.1-flash',
+    apiKey: 'k',
+    sessionId: 'session-42',
+  })
+  assert.equal(headers['x-opencode-session'], 'session-42')
+})
+
+test('buildAiChatRequest: OpenCode Go 上非 chat 协议族的模型直接给可读报错', () => {
+  assert.throws(
+    () => buildAiChatRequest({ baseURL: GO, model: 'minimax-m3', apiKey: 'k' }),
+    /minimax-m3.*Anthropic.*\/messages.*换一个/
+  )
+  assert.throws(
+    () => buildAiChatRequest({ baseURL: GO, model: 'grok-4.6', apiKey: 'k' }),
+    /grok-4\.6.*\/responses/
+  )
+})
+
+test('buildAiChatRequest: 无 apiKey 不发 Authorization，extraHeaders 可覆盖默认头', () => {
+  const { headers } = buildAiChatRequest({
+    baseURL: 'https://api.deepseek.com/v1',
+    model: 'deepseek-chat',
+    extraHeaders: { 'X-Trace': '1' },
+  })
+  assert.ok(!('Authorization' in headers))
+  assert.equal(headers['X-Trace'], '1')
+})
+
+test('buildAiChatRequest: 缺 baseURL 时明确报错（而不是拼出 /chat/completions）', () => {
+  assert.throws(() => buildAiChatRequest({ model: 'm' }), /baseURL/)
+})
+
+// ========== buildAiApiRequest ==========
+
+test('buildAiApiRequest: 拼 /models 并同样带身份头', () => {
+  const { url, headers } = buildAiApiRequest({ baseURL: GO + '/', path: '/models', apiKey: 'k' })
+  assert.equal(url, `${OPENCODE_GO_URL}/models`)
+  assert.equal(headers['Authorization'], 'Bearer k')
+  assert.ok(headers['x-opencode-session'])
+  assert.ok(headers['User-Agent'])
+})
+
+// ========== describeAiHttpError ==========
+
+test('describeAiHttpError: 抽 error.message，抽不到才退回原文', () => {
+  assert.equal(
+    describeAiHttpError(JSON.stringify({
+      error: {
+        message: 'Request is missing x-opencode-session and cannot be routed efficiently.',
+        code: 'HTTP_ERR',
+      },
+    }), 400),
+    'Request is missing x-opencode-session and cannot be routed efficiently.'
+  )
+  assert.equal(describeAiHttpError('{"error":"boom"}', 500), 'boom')
+  assert.equal(describeAiHttpError('{"message":"boom"}', 500), 'boom')
+  assert.equal(describeAiHttpError('plain failure', 500), 'plain failure')
+  assert.equal(describeAiHttpError('', 502), 'HTTP 502')
+  // 超长原文截断，不把整页 HTML 甩到界面上
+  const long = 'x'.repeat(500)
+  assert.equal(describeAiHttpError(long, 500).length, 301)
+})

package/src/ui/public/assets/AgentView-BsN9y4Kg.css

@@ -1 +0,0 @@
-.agent-view[data-v-773baa74]{background:var(--bg-container);height:100%;min-height:0;display:flex}.agent-sidebar[data-v-773baa74]{background:var(--bg-panel);border-right:1px solid var(--border-color);flex-direction:column;flex-shrink:0;display:flex;overflow:hidden}.sidebar-header[data-v-773baa74]{flex-shrink:0;padding:10px 12px}.new-session-btn[data-v-773baa74]{border:1px dashed var(--border-color);border-radius:var(--radius-md);width:100%;color:var(--text-secondary);cursor:pointer;background:0 0;justify-content:center;align-items:center;gap:6px;padding:8px 12px;font-family:inherit;font-size:13px;transition:all .15s;display:flex}.new-session-btn[data-v-773baa74]:hover{border-color:var(--color-primary);color:var(--color-primary);background:color-mix(in srgb, var(--color-primary) 6%, transparent)}.sidebar-search[data-v-773baa74]{flex-shrink:0;align-items:center;gap:6px;padding:0 12px 8px;display:flex}.sidebar-search .search-icon[data-v-773baa74]{color:var(--text-tertiary);flex-shrink:0}.sidebar-search .search-input[data-v-773baa74]{color:var(--text-primary);background:0 0;border:none;outline:none;flex:1;padding:4px 0;font-family:inherit;font-size:13px}.sidebar-search .search-input[data-v-773baa74]::placeholder{color:var(--text-tertiary)}.session-list[data-v-773baa74]{flex:1;min-height:0;padding:0 6px 8px;overflow-y:auto}.empty-state[data-v-773baa74]{text-align:center;color:var(--text-tertiary);flex-direction:column;justify-content:center;align-items:center;padding:48px 16px;display:flex}.empty-state .empty-text[data-v-773baa74]{color:var(--text-secondary);margin-top:10px;font-size:14px}.empty-state .empty-hint[data-v-773baa74]{margin-top:4px;font-size:12px}.session-item[data-v-773baa74]{border-radius:var(--radius-sm);cursor:pointer;align-items:flex-start;gap:4px;padding:8px 10px;transition:background .12s;display:flex;position:relative}.session-item[data-v-773baa74]:hover{background:var(--bg-hover)}.session-item:hover .session-item-actions[data-v-773baa74]{opacity:1}.session-item.active[data-v-773baa74]{background:color-mix(in srgb, var(--color-primary) 10%, transparent)}.session-item.active .session-item-title[data-v-773baa74]{color:var(--color-primary)}.session-item.active[data-v-773baa74]:before{content:"";background:var(--color-primary);border-radius:0 2px 2px 0;width:3px;height:20px;position:absolute;top:50%;left:0;transform:translateY(-50%)}.session-item-main[data-v-773baa74]{flex:1;min-width:0}.session-item-title[data-v-773baa74]{color:var(--text-primary);white-space:nowrap;text-overflow:ellipsis;margin-bottom:2px;font-size:13px;font-weight:500;overflow:hidden}.session-item-meta[data-v-773baa74]{color:var(--text-tertiary);align-items:center;gap:4px;font-size:11px;display:flex}.session-item-meta .meta-dot[data-v-773baa74]{opacity:.5}.session-item-meta .meta-source[data-v-773baa74]{border-radius:3px;padding:0 4px;font-size:10px;font-weight:600}.session-item-meta .meta-source.cli[data-v-773baa74]{background:color-mix(in srgb, var(--color-warning) 20%, transparent);color:var(--color-warning)}.session-item-actions[data-v-773baa74]{opacity:0;flex-shrink:0;gap:2px;transition:opacity .12s;display:flex}.item-action-btn[data-v-773baa74]{width:24px;height:24px;color:var(--text-tertiary);cursor:pointer;border-radius:var(--radius-xs);background:0 0;border:none;justify-content:center;align-items:center;padding:0;transition:all .12s;display:flex}.item-action-btn[data-v-773baa74]:hover{background:var(--bg-hover);color:var(--text-secondary)}.item-action-btn.danger[data-v-773baa74]:hover{color:var(--color-danger);background:color-mix(in srgb, var(--color-danger) 10%, transparent)}.sidebar-resizer[data-v-773baa74]{cursor:col-resize;background:0 0;flex-shrink:0;width:4px;transition:background .15s}.sidebar-resizer[data-v-773baa74]:hover,.sidebar-resizer.active[data-v-773baa74]{background:var(--color-primary);opacity:.3}.agent-chat-area[data-v-773baa74]{flex-direction:column;flex:1;min-width:0;display:flex;position:relative;overflow:hidden}.chat-loading[data-v-773baa74]{color:var(--text-tertiary);flex-direction:column;flex:1;justify-content:center;align-items:center;gap:12px;font-size:14px;display:flex}.stop-button-bar[data-v-773baa74]{z-index:10;position:absolute;bottom:80px;left:50%;transform:translate(-50%)}.stop-button[data-v-773baa74]{border:1px solid var(--color-danger);background:var(--bg-container);color:var(--color-danger);cursor:pointer;border-radius:20px;align-items:center;gap:6px;padding:6px 16px;font-family:inherit;font-size:13px;transition:all .15s;display:flex;box-shadow:0 2px 8px #0000001f}.stop-button[data-v-773baa74]:hover{background:var(--color-danger);color:#fff}.fade-enter-active[data-v-773baa74],.fade-leave-active[data-v-773baa74]{transition:opacity .2s}.fade-enter-from[data-v-773baa74],.fade-leave-to[data-v-773baa74]{opacity:0}[data-theme=dark]{background:var(--bg-panel)}