feat(antigravity): 支持 Gemini cachedContentTokenCount 映射到 Claude cache_read_input_tokens

- 在 GeminiUsageMetadata 添加 CachedContentTokenCount 字段
- 修正 token 映射：Gemini promptTokenCount 包含缓存，Claude input_tokens 不包含
- 流式和非流式响应均已支持

backend/internal/pkg/antigravity/gemini_types.go

@@ -143,9 +143,10 @@ type GeminiCandidate struct {
 
 // GeminiUsageMetadata Gemini 用量元数据
 type GeminiUsageMetadata struct {
-	PromptTokenCount     int `json:"promptTokenCount,omitempty"`
-	CandidatesTokenCount int `json:"candidatesTokenCount,omitempty"`
-	TotalTokenCount      int `json:"totalTokenCount,omitempty"`
+	PromptTokenCount        int `json:"promptTokenCount,omitempty"`
+	CandidatesTokenCount    int `json:"candidatesTokenCount,omitempty"`
+	CachedContentTokenCount int `json:"cachedContentTokenCount,omitempty"`
+	TotalTokenCount         int `json:"totalTokenCount,omitempty"`
 }
 
 // DefaultSafetySettings 默认安全设置（关闭所有过滤）

backend/internal/pkg/antigravity/response_transformer.go

@@ -232,10 +232,14 @@ func (p *NonStreamingProcessor) buildResponse(geminiResp *GeminiResponse, respon
 		stopReason = "max_tokens"
 	}
 
+	// 注意：Gemini 的 promptTokenCount 包含 cachedContentTokenCount，
+	// 但 Claude 的 input_tokens 不包含 cache_read_input_tokens，需要减去
 	usage := ClaudeUsage{}
 	if geminiResp.UsageMetadata != nil {
-		usage.InputTokens = geminiResp.UsageMetadata.PromptTokenCount
+		cached := geminiResp.UsageMetadata.CachedContentTokenCount
+		usage.InputTokens = geminiResp.UsageMetadata.PromptTokenCount - cached
 		usage.OutputTokens = geminiResp.UsageMetadata.CandidatesTokenCount
+		usage.CacheReadInputTokens = cached
 	}
 
 	// 生成响应 ID

backend/internal/pkg/antigravity/stream_transformer.go

@@ -29,8 +29,9 @@ type StreamingProcessor struct {
 	originalModel     string
 
 	// 累计 usage
-	inputTokens  int
-	outputTokens int
+	inputTokens     int
+	outputTokens    int
+	cacheReadTokens int
 }
 
 // NewStreamingProcessor 创建流式响应处理器
@@ -76,9 +77,13 @@ func (p *StreamingProcessor) ProcessLine(line string) []byte {
 	}
 
 	// 更新 usage
+	// 注意：Gemini 的 promptTokenCount 包含 cachedContentTokenCount，
+	// 但 Claude 的 input_tokens 不包含 cache_read_input_tokens，需要减去
 	if geminiResp.UsageMetadata != nil {
-		p.inputTokens = geminiResp.UsageMetadata.PromptTokenCount
+		cached := geminiResp.UsageMetadata.CachedContentTokenCount
+		p.inputTokens = geminiResp.UsageMetadata.PromptTokenCount - cached
 		p.outputTokens = geminiResp.UsageMetadata.CandidatesTokenCount
+		p.cacheReadTokens = cached
 	}
 
 	// 处理 parts
@@ -108,8 +113,9 @@ func (p *StreamingProcessor) Finish() ([]byte, *ClaudeUsage) {
 	}
 
 	usage := &ClaudeUsage{
-		InputTokens:  p.inputTokens,
-		OutputTokens: p.outputTokens,
+		InputTokens:          p.inputTokens,
+		OutputTokens:         p.outputTokens,
+		CacheReadInputTokens: p.cacheReadTokens,
 	}
 
 	return result.Bytes(), usage
@@ -123,8 +129,10 @@ func (p *StreamingProcessor) emitMessageStart(v1Resp *V1InternalResponse) []byte
 
 	usage := ClaudeUsage{}
 	if v1Resp.Response.UsageMetadata != nil {
-		usage.InputTokens = v1Resp.Response.UsageMetadata.PromptTokenCount
+		cached := v1Resp.Response.UsageMetadata.CachedContentTokenCount
+		usage.InputTokens = v1Resp.Response.UsageMetadata.PromptTokenCount - cached
 		usage.OutputTokens = v1Resp.Response.UsageMetadata.CandidatesTokenCount
+		usage.CacheReadInputTokens = cached
 	}
 
 	responseID := v1Resp.ResponseID
@@ -418,8 +426,9 @@ func (p *StreamingProcessor) emitFinish(finishReason string) []byte {
 	}
 
 	usage := ClaudeUsage{
-		InputTokens:  p.inputTokens,
-		OutputTokens: p.outputTokens,
+		InputTokens:          p.inputTokens,
+		OutputTokens:         p.outputTokens,
+		CacheReadInputTokens: p.cacheReadTokens,
 	}
 
 	deltaEvent := map[string]any{