Merge remote-tracking branch 'upstream/main'

# Conflicts:
#	src/routes/api.js

README.md

@@ -473,23 +473,23 @@ Cherry Studio支持多种AI服务的接入，下面是不同账号类型的详
 
 ```
 # API地址
-http://你的服务器:3000/claude/
+http://你的服务器:3000/claude
 
 # 模型ID示例
-claude-sonnet-4-20250514  # Claude Sonnet 4
+claude-sonnet-4-5-20250929 # Claude Sonnet 4.5
 claude-opus-4-20250514     # Claude Opus 4
 ```
 
 配置步骤：
 - 供应商类型选择"Anthropic"
-- API地址填入：`http://你的服务器:3000/claude/`
+- API地址填入：`http://你的服务器:3000/claude`
 - API Key填入：后台创建的API密钥（cr_开头）
 
 **2. Gemini账号接入：**
 
 ```
 # API地址
-http://你的服务器:3000/gemini/
+http://你的服务器:3000/gemini
 
 # 模型ID示例
 gemini-2.5-pro             # Gemini 2.5 Pro
@@ -497,14 +497,14 @@ gemini-2.5-pro             # Gemini 2.5 Pro
 
 配置步骤：
 - 供应商类型选择"Gemini"
-- API地址填入：`http://你的服务器:3000/gemini/`
+- API地址填入：`http://你的服务器:3000/gemini`
 - API Key填入：后台创建的API密钥（cr_开头）
 
 **3. Codex接入：**
 
 ```
 # API地址
-http://你的服务器:3000/openai/
+http://你的服务器:3000/openai
 
 # 模型ID（固定）
 gpt-5                      # Codex使用固定模型ID
@@ -512,10 +512,17 @@ gpt-5                      # Codex使用固定模型ID
 
 配置步骤：
 - 供应商类型选择"Openai-Response"
-- API地址填入：`http://你的服务器:3000/openai/`
+- API地址填入：`http://你的服务器:3000/openai`
 - API Key填入：后台创建的API密钥（cr_开头）
 - **重要**：Codex只支持Openai-Response标准
 
+**Cherry Studio 地址格式重要说明：**
+
+- ✅ **推荐格式**：`http://你的服务器:3000/claude`（不加结尾 `/`，让 Cherry Studio 自动加上 v1）
+- ✅ **等效格式**：`http://你的服务器:3000/claude/v1/`（手动指定 v1 并加结尾 `/`）
+- 💡 **说明**：这两种格式在 Cherry Studio 中是完全等效的
+- ❌ **错误格式**：`http://你的服务器:3000/claude/`（单独的 `/` 结尾会被 Cherry Studio 忽略 v1 版本）
+
 #### 其他第三方工具接入
 
 **接入要点：**

VERSION

@@ -1 +1 @@
-1.1.162
+1.1.164

src/routes/admin.js

@@ -4162,6 +4162,36 @@ router.get('/accounts/:accountId/usage-history', authenticateAdmin, async (req,
       gemini: 'gemini-1.5-flash'
     }
 
+    // 获取账户信息以获取创建时间
+    let accountData = null
+    let accountCreatedAt = null
+
+    try {
+      switch (platform) {
+        case 'claude':
+          accountData = await claudeAccountService.getAccount(accountId)
+          break
+        case 'claude-console':
+          accountData = await claudeConsoleAccountService.getAccount(accountId)
+          break
+        case 'openai':
+          accountData = await openaiAccountService.getAccount(accountId)
+          break
+        case 'openai-responses':
+          accountData = await openaiResponsesAccountService.getAccount(accountId)
+          break
+        case 'gemini':
+          accountData = await geminiAccountService.getAccount(accountId)
+          break
+      }
+
+      if (accountData && accountData.createdAt) {
+        accountCreatedAt = new Date(accountData.createdAt)
+      }
+    } catch (error) {
+      logger.warn(`Failed to get account data for avgDailyCost calculation: ${error.message}`)
+    }
+
     const client = redis.getClientSafe()
     const fallbackModel = fallbackModelMap[platform] || 'unknown'
     const daysCount = Math.min(Math.max(parseInt(days, 10) || 30, 1), 60)
@@ -4281,9 +4311,22 @@ router.get('/accounts/:accountId/usage-history', authenticateAdmin, async (req,
       })
     }
 
-    const avgDailyCost = daysCount > 0 ? totalCost / daysCount : 0
+    // 计算实际使用天数（从账户创建到现在）
-    const avgDailyRequests = daysCount > 0 ? totalRequests / daysCount : 0
+    let actualDaysForAvg = daysCount
-    const avgDailyTokens = daysCount > 0 ? totalTokens / daysCount : 0
+    if (accountCreatedAt) {
+      const now = new Date()
+      const diffTime = Math.abs(now - accountCreatedAt)
+      const diffDays = Math.ceil(diffTime / (1000 * 60 * 60 * 24))
+      // 使用实际使用天数，但不超过请求的天数范围
+      actualDaysForAvg = Math.min(diffDays, daysCount)
+      // 至少为1天，避免除零
+      actualDaysForAvg = Math.max(actualDaysForAvg, 1)
+    }
+
+    // 使用实际天数计算日均值
+    const avgDailyCost = actualDaysForAvg > 0 ? totalCost / actualDaysForAvg : 0
+    const avgDailyRequests = actualDaysForAvg > 0 ? totalRequests / actualDaysForAvg : 0
+    const avgDailyTokens = actualDaysForAvg > 0 ? totalTokens / actualDaysForAvg : 0
 
     const todayData = history.length > 0 ? history[history.length - 1] : null
 
@@ -4293,6 +4336,8 @@ router.get('/accounts/:accountId/usage-history', authenticateAdmin, async (req,
         history,
         summary: {
           days: daysCount,
+          actualDaysUsed: actualDaysForAvg, // 实际使用的天数（用于计算日均值）
+          accountCreatedAt: accountCreatedAt ? accountCreatedAt.toISOString() : null,
           totalCost,
           totalCostFormatted: CostCalculator.formatCost(totalCost),
           totalRequests,

src/routes/api.js

@@ -6,10 +6,8 @@ const ccrRelayService = require('../services/ccrRelayService')
 const bedrockAccountService = require('../services/bedrockAccountService')
 const unifiedClaudeScheduler = require('../services/unifiedClaudeScheduler')
 const apiKeyService = require('../services/apiKeyService')
-const pricingService = require('../services/pricingService')
 const { authenticateApiKey } = require('../middleware/auth')
 const logger = require('../utils/logger')
-const redis = require('../models/redis')
 const { getEffectiveModel, parseVendorPrefixedModel } = require('../utils/modelHelper')
 const sessionHelper = require('../utils/sessionHelper')
 const openaiToClaude = require('../services/openaiToClaude')
@@ -34,6 +32,33 @@ function detectBackendFromModel(modelName) {
   return 'claude' // 默认使用 Claude
 }
 
+const { updateRateLimitCounters } = require('../utils/rateLimitHelper')
+
+const router = express.Router()
+
+function queueRateLimitUpdate(rateLimitInfo, usageSummary, model, context = '') {
+  if (!rateLimitInfo) {
+    return Promise.resolve({ totalTokens: 0, totalCost: 0 })
+  }
+
+  const label = context ? ` (${context})` : ''
+
+  return updateRateLimitCounters(rateLimitInfo, usageSummary, model)
+    .then(({ totalTokens, totalCost }) => {
+      if (totalTokens > 0) {
+        logger.api(`📊 Updated rate limit token count${label}: +${totalTokens} tokens`)
+      }
+      if (typeof totalCost === 'number' && totalCost > 0) {
+        logger.api(`💰 Updated rate limit cost count${label}: +$${totalCost.toFixed(6)}`)
+      }
+      return { totalTokens, totalCost }
+    })
+    .catch((error) => {
+      logger.error(`❌ Failed to update rate limit counters${label}:`, error)
+      return { totalTokens: 0, totalCost: 0 }
+    })
+}
+
 // 🔧 共享的消息处理函数
 async function handleMessagesRequest(req, res) {
   try {
@@ -210,35 +235,17 @@ async function handleMessagesRequest(req, res) {
                   logger.error('❌ Failed to record stream usage:', error)
                 })
 
-              // 更新时间窗口内的token计数和费用
+              queueRateLimitUpdate(
-              if (req.rateLimitInfo) {
+                req.rateLimitInfo,
-                const totalTokens = inputTokens + outputTokens + cacheCreateTokens + cacheReadTokens
+                {
-
+                  inputTokens,
-                // 更新Token计数（向后兼容）
+                  outputTokens,
-                redis
+                  cacheCreateTokens,
-                  .getClient()
+                  cacheReadTokens
-                  .incrby(req.rateLimitInfo.tokenCountKey, totalTokens)
+                },
-                  .catch((error) => {
+                model,
-                    logger.error('❌ Failed to update rate limit token count:', error)
+                'claude-stream'
-                  })
-                logger.api(`📊 Updated rate limit token count: +${totalTokens} tokens`)
-
-                // 计算并更新费用计数（新功能）
-                if (req.rateLimitInfo.costCountKey) {
-                  const costInfo = pricingService.calculateCost(usageData, model)
-                  if (costInfo.totalCost > 0) {
-                    redis
-                      .getClient()
-                      .incrbyfloat(req.rateLimitInfo.costCountKey, costInfo.totalCost)
-                      .catch((error) => {
-                        logger.error('❌ Failed to update rate limit cost count:', error)
-                      })
-                    logger.api(
-                      `💰 Updated rate limit cost count: +$${costInfo.totalCost.toFixed(6)}`
               )
-                  }
-                }
-              }
 
               usageDataCaptured = true
               logger.api(
@@ -319,35 +326,17 @@ async function handleMessagesRequest(req, res) {
                   logger.error('❌ Failed to record stream usage:', error)
                 })
 
-              // 更新时间窗口内的token计数和费用
+              queueRateLimitUpdate(
-              if (req.rateLimitInfo) {
+                req.rateLimitInfo,
-                const totalTokens = inputTokens + outputTokens + cacheCreateTokens + cacheReadTokens
+                {
-
+                  inputTokens,
-                // 更新Token计数（向后兼容）
+                  outputTokens,
-                redis
+                  cacheCreateTokens,
-                  .getClient()
+                  cacheReadTokens
-                  .incrby(req.rateLimitInfo.tokenCountKey, totalTokens)
+                },
-                  .catch((error) => {
+                model,
-                    logger.error('❌ Failed to update rate limit token count:', error)
+                'claude-console-stream'
-                  })
-                logger.api(`📊 Updated rate limit token count: +${totalTokens} tokens`)
-
-                // 计算并更新费用计数（新功能）
-                if (req.rateLimitInfo.costCountKey) {
-                  const costInfo = pricingService.calculateCost(usageData, model)
-                  if (costInfo.totalCost > 0) {
-                    redis
-                      .getClient()
-                      .incrbyfloat(req.rateLimitInfo.costCountKey, costInfo.totalCost)
-                      .catch((error) => {
-                        logger.error('❌ Failed to update rate limit cost count:', error)
-                      })
-                    logger.api(
-                      `💰 Updated rate limit cost count: +$${costInfo.totalCost.toFixed(6)}`
               )
-                  }
-                }
-              }
 
               usageDataCaptured = true
               logger.api(
@@ -387,33 +376,17 @@ async function handleMessagesRequest(req, res) {
                 logger.error('❌ Failed to record Bedrock stream usage:', error)
               })
 
-            // 更新时间窗口内的token计数和费用
+            queueRateLimitUpdate(
-            if (req.rateLimitInfo) {
+              req.rateLimitInfo,
-              const totalTokens = inputTokens + outputTokens
+              {
-
+                inputTokens,
-              // 更新Token计数（向后兼容）
+                outputTokens,
-              redis
+                cacheCreateTokens: 0,
-                .getClient()
+                cacheReadTokens: 0
-                .incrby(req.rateLimitInfo.tokenCountKey, totalTokens)
+              },
-                .catch((error) => {
+              result.model,
-                  logger.error('❌ Failed to update rate limit token count:', error)
+              'bedrock-stream'
-                })
+            )
-              logger.api(`📊 Updated rate limit token count: +${totalTokens} tokens`)
-
-              // 计算并更新费用计数（新功能）
-              if (req.rateLimitInfo.costCountKey) {
-                const costInfo = pricingService.calculateCost(result.usage, result.model)
-                if (costInfo.totalCost > 0) {
-                  redis
-                    .getClient()
-                    .incrbyfloat(req.rateLimitInfo.costCountKey, costInfo.totalCost)
-                    .catch((error) => {
-                      logger.error('❌ Failed to update rate limit cost count:', error)
-                    })
-                  logger.api(`💰 Updated rate limit cost count: +$${costInfo.totalCost.toFixed(6)}`)
-                }
-              }
-            }
 
             usageDataCaptured = true
             logger.api(
@@ -488,35 +461,17 @@ async function handleMessagesRequest(req, res) {
                   logger.error('❌ Failed to record CCR stream usage:', error)
                 })
 
-              // 更新时间窗口内的token计数和费用
+              queueRateLimitUpdate(
-              if (req.rateLimitInfo) {
+                req.rateLimitInfo,
-                const totalTokens = inputTokens + outputTokens + cacheCreateTokens + cacheReadTokens
+                {
-
+                  inputTokens,
-                // 更新Token计数（向后兼容）
+                  outputTokens,
-                redis
+                  cacheCreateTokens,
-                  .getClient()
+                  cacheReadTokens
-                  .incrby(req.rateLimitInfo.tokenCountKey, totalTokens)
+                },
-                  .catch((error) => {
+                model,
-                    logger.error('❌ Failed to update rate limit token count:', error)
+                'ccr-stream'
-                  })
-                logger.api(`📊 Updated rate limit token count: +${totalTokens} tokens`)
-
-                // 计算并更新费用计数（新功能）
-                if (req.rateLimitInfo.costCountKey) {
-                  const costInfo = pricingService.calculateCost(usageData, model)
-                  if (costInfo.totalCost > 0) {
-                    redis
-                      .getClient()
-                      .incrbyfloat(req.rateLimitInfo.costCountKey, costInfo.totalCost)
-                      .catch((error) => {
-                        logger.error('❌ Failed to update rate limit cost count:', error)
-                      })
-                    logger.api(
-                      `💰 Updated rate limit cost count: +$${costInfo.totalCost.toFixed(6)}`
               )
-                  }
-                }
-              }
 
               usageDataCaptured = true
               logger.api(
@@ -704,25 +659,17 @@ async function handleMessagesRequest(req, res) {
             responseAccountId
           )
 
-          // 更新时间窗口内的token计数和费用
+          await queueRateLimitUpdate(
-          if (req.rateLimitInfo) {
+            req.rateLimitInfo,
-            const totalTokens = inputTokens + outputTokens + cacheCreateTokens + cacheReadTokens
+            {
-
+              inputTokens,
-            // 更新Token计数（向后兼容）
+              outputTokens,
-            await redis.getClient().incrby(req.rateLimitInfo.tokenCountKey, totalTokens)
+              cacheCreateTokens,
-            logger.api(`📊 Updated rate limit token count: +${totalTokens} tokens`)
+              cacheReadTokens
-
+            },
-            // 计算并更新费用计数（新功能）
+            model,
-            if (req.rateLimitInfo.costCountKey) {
+            'claude-non-stream'
-              const costInfo = pricingService.calculateCost(jsonData.usage, model)
+          )
-              if (costInfo.totalCost > 0) {
-                await redis
-                  .getClient()
-                  .incrbyfloat(req.rateLimitInfo.costCountKey, costInfo.totalCost)
-                logger.api(`💰 Updated rate limit cost count: +$${costInfo.totalCost.toFixed(6)}`)
-              }
-            }
-          }
 
           usageRecorded = true
           logger.api(

src/routes/geminiRoutes.js

@@ -8,6 +8,7 @@ const crypto = require('crypto')
 const sessionHelper = require('../utils/sessionHelper')
 const unifiedGeminiScheduler = require('../services/unifiedGeminiScheduler')
 const apiKeyService = require('../services/apiKeyService')
+const { updateRateLimitCounters } = require('../utils/rateLimitHelper')
 // const { OAuth2Client } = require('google-auth-library'); // OAuth2Client is not used in this file
 
 // 生成会话哈希
@@ -49,6 +50,31 @@ function ensureGeminiPermission(req, res) {
   return false
 }
 
+async function applyRateLimitTracking(req, usageSummary, model, context = '') {
+  if (!req.rateLimitInfo) {
+    return
+  }
+
+  const label = context ? ` (${context})` : ''
+
+  try {
+    const { totalTokens, totalCost } = await updateRateLimitCounters(
+      req.rateLimitInfo,
+      usageSummary,
+      model
+    )
+
+    if (totalTokens > 0) {
+      logger.api(`📊 Updated rate limit token count${label}: +${totalTokens} tokens`)
+    }
+    if (typeof totalCost === 'number' && totalCost > 0) {
+      logger.api(`💰 Updated rate limit cost count${label}: +$${totalCost.toFixed(6)}`)
+    }
+  } catch (error) {
+    logger.error(`❌ Failed to update rate limit counters${label}:`, error)
+  }
+}
+
 // Gemini 消息处理端点
 router.post('/messages', authenticateApiKey, async (req, res) => {
   const startTime = Date.now()
@@ -679,6 +705,18 @@ async function handleGenerateContent(req, res) {
         logger.info(
           `📊 Recorded Gemini usage - Input: ${usage.promptTokenCount}, Output: ${usage.candidatesTokenCount}, Total: ${usage.totalTokenCount}`
         )
+
+        await applyRateLimitTracking(
+          req,
+          {
+            inputTokens: usage.promptTokenCount || 0,
+            outputTokens: usage.candidatesTokenCount || 0,
+            cacheCreateTokens: 0,
+            cacheReadTokens: 0
+          },
+          model,
+          'gemini-non-stream'
+        )
       } catch (error) {
         logger.error('Failed to record Gemini usage:', error)
       }
@@ -935,6 +973,18 @@ async function handleStreamGenerateContent(req, res) {
           logger.info(
             `📊 Recorded Gemini stream usage - Input: ${totalUsage.promptTokenCount}, Output: ${totalUsage.candidatesTokenCount}, Total: ${totalUsage.totalTokenCount}`
           )
+
+          await applyRateLimitTracking(
+            req,
+            {
+              inputTokens: totalUsage.promptTokenCount || 0,
+              outputTokens: totalUsage.candidatesTokenCount || 0,
+              cacheCreateTokens: 0,
+              cacheReadTokens: 0
+            },
+            model,
+            'gemini-stream'
+          )
         } catch (error) {
           logger.error('Failed to record Gemini usage:', error)
         }

src/routes/openaiClaudeRoutes.js

@@ -15,6 +15,7 @@ const apiKeyService = require('../services/apiKeyService')
 const unifiedClaudeScheduler = require('../services/unifiedClaudeScheduler')
 const claudeCodeHeadersService = require('../services/claudeCodeHeadersService')
 const sessionHelper = require('../utils/sessionHelper')
+const { updateRateLimitCounters } = require('../utils/rateLimitHelper')
 
 // 加载模型定价数据
 let modelPricingData = {}
@@ -33,6 +34,27 @@ function checkPermissions(apiKeyData, requiredPermission = 'claude') {
   return permissions === 'all' || permissions === requiredPermission
 }
 
+function queueRateLimitUpdate(rateLimitInfo, usageSummary, model, context = '') {
+  if (!rateLimitInfo) {
+    return
+  }
+
+  const label = context ? ` (${context})` : ''
+
+  updateRateLimitCounters(rateLimitInfo, usageSummary, model)
+    .then(({ totalTokens, totalCost }) => {
+      if (totalTokens > 0) {
+        logger.api(`📊 Updated rate limit token count${label}: +${totalTokens} tokens`)
+      }
+      if (typeof totalCost === 'number' && totalCost > 0) {
+        logger.api(`💰 Updated rate limit cost count${label}: +$${totalCost.toFixed(6)}`)
+      }
+    })
+    .catch((error) => {
+      logger.error(`❌ Failed to update rate limit counters${label}:`, error)
+    })
+}
+
 // 📋 OpenAI 兼容的模型列表端点
 router.get('/v1/models', authenticateApiKey, async (req, res) => {
   try {
@@ -263,6 +285,12 @@ async function handleChatCompletion(req, res, apiKeyData) {
           // 记录使用统计
           if (usage && usage.input_tokens !== undefined && usage.output_tokens !== undefined) {
             const model = usage.model || claudeRequest.model
+            const cacheCreateTokens =
+              (usage.cache_creation && typeof usage.cache_creation === 'object'
+                ? (usage.cache_creation.ephemeral_5m_input_tokens || 0) +
+                  (usage.cache_creation.ephemeral_1h_input_tokens || 0)
+                : usage.cache_creation_input_tokens || 0) || 0
+            const cacheReadTokens = usage.cache_read_input_tokens || 0
 
             // 使用新的 recordUsageWithDetails 方法来支持详细的缓存数据
             apiKeyService
@@ -275,6 +303,18 @@ async function handleChatCompletion(req, res, apiKeyData) {
               .catch((error) => {
                 logger.error('❌ Failed to record usage:', error)
               })
+
+            queueRateLimitUpdate(
+              req.rateLimitInfo,
+              {
+                inputTokens: usage.input_tokens || 0,
+                outputTokens: usage.output_tokens || 0,
+                cacheCreateTokens,
+                cacheReadTokens
+              },
+              model,
+              'openai-claude-stream'
+            )
           }
         },
         // 流转换器
@@ -334,6 +374,12 @@ async function handleChatCompletion(req, res, apiKeyData) {
       // 记录使用统计
       if (claudeData.usage) {
         const { usage } = claudeData
+        const cacheCreateTokens =
+          (usage.cache_creation && typeof usage.cache_creation === 'object'
+            ? (usage.cache_creation.ephemeral_5m_input_tokens || 0) +
+              (usage.cache_creation.ephemeral_1h_input_tokens || 0)
+            : usage.cache_creation_input_tokens || 0) || 0
+        const cacheReadTokens = usage.cache_read_input_tokens || 0
         // 使用新的 recordUsageWithDetails 方法来支持详细的缓存数据
         apiKeyService
           .recordUsageWithDetails(
@@ -345,6 +391,18 @@ async function handleChatCompletion(req, res, apiKeyData) {
           .catch((error) => {
             logger.error('❌ Failed to record usage:', error)
           })
+
+        queueRateLimitUpdate(
+          req.rateLimitInfo,
+          {
+            inputTokens: usage.input_tokens || 0,
+            outputTokens: usage.output_tokens || 0,
+            cacheCreateTokens,
+            cacheReadTokens
+          },
+          claudeRequest.model,
+          'openai-claude-non-stream'
+        )
       }
 
       // 返回 OpenAI 格式响应

src/routes/openaiRoutes.js

@@ -11,6 +11,7 @@ const openaiResponsesRelayService = require('../services/openaiResponsesRelaySer
 const apiKeyService = require('../services/apiKeyService')
 const crypto = require('crypto')
 const ProxyHelper = require('../utils/proxyHelper')
+const { updateRateLimitCounters } = require('../utils/rateLimitHelper')
 
 // 创建代理 Agent（使用统一的代理工具）
 function createProxyAgent(proxy) {
@@ -67,6 +68,31 @@ function extractCodexUsageHeaders(headers) {
   return hasData ? snapshot : null
 }
 
+async function applyRateLimitTracking(req, usageSummary, model, context = '') {
+  if (!req.rateLimitInfo) {
+    return
+  }
+
+  const label = context ? ` (${context})` : ''
+
+  try {
+    const { totalTokens, totalCost } = await updateRateLimitCounters(
+      req.rateLimitInfo,
+      usageSummary,
+      model
+    )
+
+    if (totalTokens > 0) {
+      logger.api(`📊 Updated rate limit token count${label}: +${totalTokens} tokens`)
+    }
+    if (typeof totalCost === 'number' && totalCost > 0) {
+      logger.api(`💰 Updated rate limit cost count${label}: +$${totalCost.toFixed(6)}`)
+    }
+  } catch (error) {
+    logger.error(`❌ Failed to update rate limit counters${label}:`, error)
+  }
+}
+
 // 使用统一调度器选择 OpenAI 账户
 async function getOpenAIAuthToken(apiKeyData, sessionId = null, requestedModel = null) {
   try {
@@ -579,6 +605,18 @@ const handleResponses = async (req, res) => {
           logger.info(
             `📊 Recorded OpenAI non-stream usage - Input: ${totalInputTokens}(actual:${actualInputTokens}+cached:${cacheReadTokens}), Output: ${outputTokens}, Total: ${usageData.total_tokens || totalInputTokens + outputTokens}, Model: ${actualModel}`
           )
+
+          await applyRateLimitTracking(
+            req,
+            {
+              inputTokens: actualInputTokens,
+              outputTokens,
+              cacheCreateTokens: 0,
+              cacheReadTokens
+            },
+            actualModel,
+            'openai-non-stream'
+          )
         }
 
         // 返回响应
@@ -700,6 +738,18 @@ const handleResponses = async (req, res) => {
             `📊 Recorded OpenAI usage - Input: ${totalInputTokens}(actual:${actualInputTokens}+cached:${cacheReadTokens}), Output: ${outputTokens}, Total: ${usageData.total_tokens || totalInputTokens + outputTokens}, Model: ${modelToRecord} (actual: ${actualModel}, requested: ${requestedModel})`
           )
           usageReported = true
+
+          await applyRateLimitTracking(
+            req,
+            {
+              inputTokens: actualInputTokens,
+              outputTokens,
+              cacheCreateTokens: 0,
+              cacheReadTokens
+            },
+            modelToRecord,
+            'openai-stream'
+          )
         } catch (error) {
           logger.error('Failed to record OpenAI usage:', error)
         }

src/services/claudeRelayService.js

@@ -39,17 +39,8 @@ class ClaudeRelayService {
   }
 
   // 🔍 判断是否是真实的 Claude Code 请求
-  isRealClaudeCodeRequest(requestBody, clientHeaders) {
+  isRealClaudeCodeRequest(requestBody) {
-    // 使用 claudeCodeValidator 来进行完整的验证
+    return ClaudeCodeValidator.hasClaudeCodeSystemPrompt(requestBody)
-    // 注意：claudeCodeValidator.validate() 需要一个完整的 req 对象
-    // 我们需要构造一个最小化的 req 对象来满足验证器的需求
-    const mockReq = {
-      headers: clientHeaders || {},
-      body: requestBody,
-      path: '/api/v1/messages'
-    }
-
-    return ClaudeCodeValidator.validate(mockReq)
   }
 
   // 🚀 转发请求到Claude API
@@ -151,8 +142,7 @@ class ClaudeRelayService {
       // 获取有效的访问token
       const accessToken = await claudeAccountService.getValidAccessToken(accountId)
 
-      // 处理请求体（传递 clientHeaders 以判断是否需要设置 Claude Code 系统提示词）
+      const processedBody = this._processRequestBody(requestBody, account)
-      const processedBody = this._processRequestBody(requestBody, clientHeaders, account)
 
       // 获取代理配置
       const proxyAgent = await this._getProxyAgent(accountId)
@@ -397,7 +387,7 @@ class ClaudeRelayService {
         if (
           clientHeaders &&
           Object.keys(clientHeaders).length > 0 &&
-          this.isRealClaudeCodeRequest(requestBody, clientHeaders)
+          this.isRealClaudeCodeRequest(requestBody)
         ) {
           await claudeCodeHeadersService.storeAccountHeaders(accountId, clientHeaders)
         }
@@ -444,7 +434,7 @@ class ClaudeRelayService {
   }
 
   // 🔄 处理请求体
-  _processRequestBody(body, clientHeaders = {}, account = null) {
+  _processRequestBody(body, account = null) {
     if (!body) {
       return body
     }
@@ -459,7 +449,7 @@ class ClaudeRelayService {
     this._stripTtlFromCacheControl(processedBody)
 
     // 判断是否是真实的 Claude Code 请求
-    const isRealClaudeCode = this.isRealClaudeCodeRequest(processedBody, clientHeaders)
+    const isRealClaudeCode = this.isRealClaudeCodeRequest(processedBody)
 
     // 如果不是真实的 Claude Code 请求，需要设置 Claude Code 系统提示词
     if (!isRealClaudeCode) {
@@ -760,7 +750,7 @@ class ClaudeRelayService {
     const filteredHeaders = this._filterClientHeaders(clientHeaders)
 
     // 判断是否是真实的 Claude Code 请求
-    const isRealClaudeCode = this.isRealClaudeCodeRequest(body, clientHeaders)
+    const isRealClaudeCode = this.isRealClaudeCodeRequest(body)
 
     // 如果不是真实的 Claude Code 请求，需要使用从账户获取的 Claude Code headers
     const finalHeaders = { ...filteredHeaders }
@@ -1007,8 +997,7 @@ class ClaudeRelayService {
       // 获取有效的访问token
       const accessToken = await claudeAccountService.getValidAccessToken(accountId)
 
-      // 处理请求体（传递 clientHeaders 以判断是否需要设置 Claude Code 系统提示词）
+      const processedBody = this._processRequestBody(requestBody, account)
-      const processedBody = this._processRequestBody(requestBody, clientHeaders, account)
 
       // 获取代理配置
       const proxyAgent = await this._getProxyAgent(accountId)
@@ -1065,7 +1054,7 @@ class ClaudeRelayService {
     const filteredHeaders = this._filterClientHeaders(clientHeaders)
 
     // 判断是否是真实的 Claude Code 请求
-    const isRealClaudeCode = this.isRealClaudeCodeRequest(body, clientHeaders)
+    const isRealClaudeCode = this.isRealClaudeCodeRequest(body)
 
     // 如果不是真实的 Claude Code 请求，需要使用从账户获取的 Claude Code headers
     const finalHeaders = { ...filteredHeaders }
@@ -1595,7 +1584,7 @@ class ClaudeRelayService {
             if (
               clientHeaders &&
               Object.keys(clientHeaders).length > 0 &&
-              this.isRealClaudeCodeRequest(body, clientHeaders)
+              this.isRealClaudeCodeRequest(body)
             ) {
               await claudeCodeHeadersService.storeAccountHeaders(accountId, clientHeaders)
             }

src/utils/contents.js

@@ -74,6 +74,11 @@ const PROMPT_DEFINITIONS = {
     title: 'Claude Agent SDK System Prompt',
     text: "You are a Claude agent, built on Anthropic's Claude Agent SDK."
   },
+  claudeOtherSystemPrompt4: {
+    category: 'system',
+    title: 'Claude Code Compact System Prompt Agent SDK2',
+    text: "You are Claude Code, Anthropic's official CLI for Claude, running within the Claude Agent SDK."
+  },
   claudeOtherSystemPromptCompact: {
     category: 'system',
     title: 'Claude Code Compact System Prompt',

src/utils/rateLimitHelper.js

@@ -0,0 +1,71 @@
+const redis = require('../models/redis')
+const pricingService = require('../services/pricingService')
+const CostCalculator = require('./costCalculator')
+
+function toNumber(value) {
+  const num = Number(value)
+  return Number.isFinite(num) ? num : 0
+}
+
+async function updateRateLimitCounters(rateLimitInfo, usageSummary, model) {
+  if (!rateLimitInfo) {
+    return { totalTokens: 0, totalCost: 0 }
+  }
+
+  const client = redis.getClient()
+  if (!client) {
+    throw new Error('Redis 未连接，无法更新限流计数')
+  }
+
+  const inputTokens = toNumber(usageSummary.inputTokens)
+  const outputTokens = toNumber(usageSummary.outputTokens)
+  const cacheCreateTokens = toNumber(usageSummary.cacheCreateTokens)
+  const cacheReadTokens = toNumber(usageSummary.cacheReadTokens)
+
+  const totalTokens = inputTokens + outputTokens + cacheCreateTokens + cacheReadTokens
+
+  if (totalTokens > 0 && rateLimitInfo.tokenCountKey) {
+    await client.incrby(rateLimitInfo.tokenCountKey, Math.round(totalTokens))
+  }
+
+  let totalCost = 0
+  const usagePayload = {
+    input_tokens: inputTokens,
+    output_tokens: outputTokens,
+    cache_creation_input_tokens: cacheCreateTokens,
+    cache_read_input_tokens: cacheReadTokens
+  }
+
+  try {
+    const costInfo = pricingService.calculateCost(usagePayload, model)
+    const { totalCost: calculatedCost } = costInfo || {}
+    if (typeof calculatedCost === 'number') {
+      totalCost = calculatedCost
+    }
+  } catch (error) {
+    // 忽略此处错误，后续使用备用计算
+    totalCost = 0
+  }
+
+  if (totalCost === 0) {
+    try {
+      const fallback = CostCalculator.calculateCost(usagePayload, model)
+      const { costs } = fallback || {}
+      if (costs && typeof costs.total === 'number') {
+        totalCost = costs.total
+      }
+    } catch (error) {
+      totalCost = 0
+    }
+  }
+
+  if (totalCost > 0 && rateLimitInfo.costCountKey) {
+    await client.incrbyfloat(rateLimitInfo.costCountKey, totalCost)
+  }
+
+  return { totalTokens, totalCost }
+}
+
+module.exports = {
+  updateRateLimitCounters
+}

src/validators/clients/claudeCodeValidator.js

@@ -74,16 +74,7 @@ class ClaudeCodeValidator {
       const userAgent = req.headers['user-agent'] || ''
       const path = req.path || ''
 
-      // 1. 先检查是否是 Claude Code 的 User-Agent
+      const claudeCodePattern = /^claude-cli\/\d+\.\d+\.\d+/i;
-      // 支持的格式:
-      // - claude-cli/1.0.86 (external, cli) - 原有 CLI 格式
-      // - claude-cli/2.0.0 (external, claude-vscode) - VSCode 插件格式
-      // - claude-cli/x.x.x (external, sdk-py) - Python SDK 格式
-      // - claude-cli/x.x.x (external, sdk-js) - JavaScript SDK 格式
-      // - 其他 (external, claude-xxx) 或 (external, sdk-xxx) 格式
-
-      const claudeCodePattern =
-        /^claude-cli\/[\d.]+(?:[-\w]*)?\s+\(external,\s*(?:cli|claude-[\w-]+|sdk-[\w-]+)\)$/i
 
       if (!claudeCodePattern.test(userAgent)) {
         // 不是 Claude Code 的请求，此验证器不处理

web/admin-spa/src/components/accounts/AccountUsageDetailModal.vue

@@ -38,6 +38,9 @@
               </div>
               <p class="text-xs text-gray-500 dark:text-gray-400 sm:text-sm">
                 近 {{ summary?.days || 30 }} 天内的费用与请求趋势
+                <span v-if="summary?.actualDaysUsed && summary?.actualDaysUsed < summary?.days">
+                  (日均基于实际使用 {{ summary.actualDaysUsed }} 天)
+                </span>
               </p>
             </div>
           </div>
@@ -443,7 +446,10 @@ const primaryMetrics = computed(() => [
     key: 'avgCost',
     label: '日均费用',
     value: props.summary?.avgDailyCostFormatted || formatCost(props.summary?.avgDailyCost || 0),
-    subtitle: '平均每日成本',
+    subtitle:
+      props.summary?.actualDaysUsed && props.summary?.actualDaysUsed < props.summary?.days
+        ? `基于 ${props.summary.actualDaysUsed} 天实际使用`
+        : '平均每日成本',
     icon: 'fa-wave-square',
     iconClass: 'text-purple-500'
   },

web/admin-spa/src/stores/user.js

@@ -1,8 +1,9 @@
 import { defineStore } from 'pinia'
 import axios from 'axios'
 import { showToast } from '@/utils/toast'
+import { API_PREFIX } from '@/config/api'
 
-const API_BASE = '/users'
+const API_BASE = `${API_PREFIX}/users`
 
 export const useUserStore = defineStore('user', {
   state: () => ({