feat(relay): hosted-tool conversion fidelity, reasoning normalization, and billing usage integrity (#7137)

* feat(relaykit): preserve hosted tools across conversions

- add protocol-neutral hosted-tool DTOs, conversion metadata, and loss policies
- bridge citations, grounding metadata, and hosted-tool stream lifecycles
- document the public conversion behavior and channel policy controls

* refactor(relaykit): normalize reasoning and thinking intent

- centralize provider-neutral reasoning intent, effort, and budget mappings
- parse model suffixes at the host entry boundary while preserving provider-owned tails
- keep adaptive Claude thinking and explicit zero-token compatibility consistent

* fix(billing): preserve authoritative usage across relay hops

- carry native BillingUsage sidecars through direct and streamed protocol bridges
- merge partial and terminal usage monotonically with safe fallback settlement
- retain cache metadata, penultimate usage, and per-call Gemini tool surcharges

* feat(relay): bridge Responses with Claude and Gemini protocols

- add direct request, response, and stream converters across supported relay formats
- expose Claude count_tokens and Chat-to-Responses compatibility endpoints
- carry conversion diagnostics through the host while retaining the curated public goldens

* fix(relay): wire relaykit conversions into host channels

- connect handlers, adaptors, and channel settings to the standalone conversion layer
- keep model mapping, pricing identity, retries, and provider-specific suffix behavior aligned
- ignore local audit artifacts and retain focused public regression coverage

controller/relay.go

@@ -238,7 +238,7 @@ func Relay(c *gin.Context, relayFormat types.RelayFormat) {
 		newAPIError = service.NormalizeViolationFeeError(newAPIError)
 		relayInfo.LastError = newAPIError
 
-		processChannelError(c, *types.NewChannelError(channel.Id, channel.Type, channel.Name, channel.ChannelInfo.IsMultiKey, common.GetContextKeyString(c, constant.ContextKeyChannelKey), channel.GetAutoBan()), newAPIError)
+		processChannelError(c, *types.NewChannelError(channel.Id, channel.Type, channel.Name, channel.ChannelInfo.IsMultiKey, common.GetContextKeyString(c, constant.ContextKeyChannelKey), channel.GetAutoBan()), newAPIError, relayInfo)
 
 		if !shouldRetry(c, newAPIError, common.RetryTimes-retryParam.GetRetry()) {
 			break
@@ -257,6 +257,38 @@ func Relay(c *gin.Context, relayFormat types.RelayFormat) {
 	}
 }
 
+// CountClaudeTokens implements Anthropic's token-counting utility endpoint.
+// It deliberately skips upstream generation and billing; callers use this
+// endpoint to size prompts before creating a Message.
+func CountClaudeTokens(c *gin.Context) {
+	request, err := helper.GetAndValidateClaudeRequest(c)
+	if err != nil {
+		c.JSON(http.StatusBadRequest, gin.H{
+			"type": "error",
+			"error": gin.H{
+				"type":    "invalid_request_error",
+				"message": common.MessageWithRequestId(err.Error(), c.GetString(common.RequestIdKey)),
+			},
+		})
+		return
+	}
+
+	info := relaycommon.GenRelayInfoClaude(c, request)
+	inputTokens, err := service.CountRequestToken(c, request.GetTokenCountMeta(), info)
+	if err != nil {
+		c.JSON(http.StatusInternalServerError, gin.H{
+			"type": "error",
+			"error": gin.H{
+				"type":    "api_error",
+				"message": common.MessageWithRequestId(err.Error(), c.GetString(common.RequestIdKey)),
+			},
+		})
+		return
+	}
+
+	c.JSON(http.StatusOK, gin.H{"input_tokens": inputTokens})
+}
+
 var upgrader = websocket.Upgrader{
 	Subprotocols: []string{"realtime"}, // WS 握手支持的协议，如果有使用 Sec-WebSocket-Protocol，则必须在此声明对应的 Protocol TODO add other protocol
 	CheckOrigin: func(r *http.Request) bool {
@@ -362,7 +394,7 @@ func shouldRetry(c *gin.Context, openaiErr *types.NewAPIError, retryTimes int) b
 	return operation_setting.ShouldRetryByStatusCode(code)
 }
 
-func processChannelError(c *gin.Context, channelError types.ChannelError, err *types.NewAPIError) {
+func processChannelError(c *gin.Context, channelError types.ChannelError, err *types.NewAPIError, relayInfo *relaycommon.RelayInfo) {
 	logger.LogError(c, fmt.Sprintf("channel error (channel #%d, status code: %d): %s", channelError.ChannelId, err.StatusCode, common.LocalLogPreview(err.Error())))
 	// 不要使用context获取渠道信息，异步处理时可能会出现渠道信息不一致的情况
 	// do not use context to get channel info, there may be inconsistent channel info when processing asynchronously
@@ -392,6 +424,14 @@ func processChannelError(c *gin.Context, channelError types.ChannelError, err *t
 		other["channel_type"] = c.GetInt("channel_type")
 		adminInfo := make(map[string]interface{})
 		adminInfo["use_channel"] = c.GetStringSlice("use_channel")
+		if relayInfo != nil {
+			if diagnostics := relayInfo.ConversionDiagnostics(); len(diagnostics) > 0 {
+				adminInfo["conversion_diagnostics"] = diagnostics
+			}
+			if relayInfo.ConversionDiagnosticsTruncated() {
+				adminInfo["conversion_diagnostics_truncated"] = true
+			}
+		}
 		isMultiKey := common.GetContextKeyBool(c, constant.ContextKeyChannelIsMultiKey)
 		if isMultiKey {
 			adminInfo["is_multi_key"] = true
@@ -655,7 +695,8 @@ func executeTaskSubmissionWith(
 			processChannelError(c,
 				*types.NewChannelError(channel.Id, channel.Type, channel.Name, channel.ChannelInfo.IsMultiKey,
 					common.GetContextKeyString(c, constant.ContextKeyChannelKey), channel.GetAutoBan()),
-				types.NewOpenAIError(taskErr.Error, types.ErrorCodeBadResponseStatusCode, taskErr.StatusCode))
+				types.NewOpenAIError(taskErr.Error, types.ErrorCodeBadResponseStatusCode, taskErr.StatusCode),
+				relayInfo)
 		}
 
 		willRetry := shouldRetryTaskRelay(c, channel.Id, taskErr, common.RetryTimes-retryParam.GetRetry())