fix: use totalUsage with all token types for accurate quota tracking

The onFinish callback's 'usage' only contains the final step's tokens,
which underreports usage for multi-step tool calls (like diagram generation).
Changed to 'totalUsage' which provides cumulative counts across all steps.

Include all 4 token types for accurate counting:
1. inputTokens - non-cached input tokens
2. outputTokens - generated output tokens
3. cachedInputTokens - tokens read from prompt cache
4. inputTokenDetails.cacheWriteTokens - tokens written to cache

Tested locally:
- Request 1 (cache write): 334 + 62 + 0 + 6671 = 7,067 tokens
- Request 2 (cache read): 334 + 184 + 6551 + 120 = 7,189 tokens
- DynamoDB total: 14,256 ✓

app/api/chat/route.ts

@@ -14,6 +14,11 @@ import path from "path"
 import { z } from "zod"
 import { getAIModel, supportsPromptCaching } from "@/lib/ai-providers"
 import { findCachedResponse } from "@/lib/cached-responses"
+import {
+    checkAndIncrementRequest,
+    isQuotaEnabled,
+    recordTokenUsage,
+} from "@/lib/dynamo-quota-manager"
 import {
     getTelemetryConfig,
     setTraceInput,
@@ -162,9 +167,13 @@ async function handleChatRequest(req: Request): Promise<Response> {
 
     const { messages, xml, previousXml, sessionId } = await req.json()
 
-    // Get user IP for Langfuse tracking
+    // Get user IP for Langfuse tracking (hashed for privacy)
     const forwardedFor = req.headers.get("x-forwarded-for")
-    const userId = forwardedFor?.split(",")[0]?.trim() || "anonymous"
+    const rawIp = forwardedFor?.split(",")[0]?.trim() || "anonymous"
+    const userId =
+        rawIp === "anonymous"
+            ? rawIp
+            : `user-${Buffer.from(rawIp).toString("base64url").slice(0, 8)}`
 
     // Validate sessionId for Langfuse (must be string, max 200 chars)
     const validSessionId =
@@ -187,6 +196,33 @@ async function handleChatRequest(req: Request): Promise<Response> {
         userId: userId,
     })
 
+    // === SERVER-SIDE QUOTA CHECK START ===
+    // Quota is opt-in: only enabled when DYNAMODB_QUOTA_TABLE env var is set
+    const hasOwnApiKey = !!(
+        req.headers.get("x-ai-provider") && req.headers.get("x-ai-api-key")
+    )
+
+    // Skip quota check if: quota disabled, user has own API key, or is anonymous
+    if (isQuotaEnabled() && !hasOwnApiKey && userId !== "anonymous") {
+        const quotaCheck = await checkAndIncrementRequest(userId, {
+            requests: Number(process.env.DAILY_REQUEST_LIMIT) || 10,
+            tokens: Number(process.env.DAILY_TOKEN_LIMIT) || 200000,
+            tpm: Number(process.env.TPM_LIMIT) || 20000,
+        })
+        if (!quotaCheck.allowed) {
+            return Response.json(
+                {
+                    error: quotaCheck.error,
+                    type: quotaCheck.type,
+                    used: quotaCheck.used,
+                    limit: quotaCheck.limit,
+                },
+                { status: 429 },
+            )
+        }
+    }
+    // === SERVER-SIDE QUOTA CHECK END ===
+
     // === FILE VALIDATION START ===
     const fileValidation = validateFileParts(messages)
     if (!fileValidation.valid) {
@@ -506,12 +542,26 @@ ${userInputText}
                 userId,
             }),
         }),
-        onFinish: ({ text, usage }) => {
-            // Pass usage to Langfuse (Bedrock streaming doesn't auto-report tokens to telemetry)
-            setTraceOutput(text, {
-                promptTokens: usage?.inputTokens,
-                completionTokens: usage?.outputTokens,
-            })
+        onFinish: ({ text, totalUsage }) => {
+            // AI SDK 6 telemetry auto-reports token usage on its spans
+            setTraceOutput(text)
+
+            // Record token usage for server-side quota tracking (if enabled)
+            // Use totalUsage (cumulative across all steps) instead of usage (final step only)
+            // Include all 4 token types: input, output, cache read, cache write
+            if (
+                isQuotaEnabled() &&
+                !hasOwnApiKey &&
+                userId !== "anonymous" &&
+                totalUsage
+            ) {
+                const totalTokens =
+                    (totalUsage.inputTokens || 0) +
+                    (totalUsage.outputTokens || 0) +
+                    (totalUsage.cachedInputTokens || 0) +
+                    (totalUsage.inputTokenDetails?.cacheWriteTokens || 0)
+                recordTokenUsage(userId, totalTokens)
+            }
         },
         tools: {
             // Client-side tool that will be executed on the client
@@ -681,20 +731,9 @@ Call this tool to get shape names and usage syntax for a specific library.`,
         messageMetadata: ({ part }) => {
             if (part.type === "finish") {
                 const usage = (part as any).totalUsage
-                if (!usage) {
-                    console.warn(
-                        "[messageMetadata] No usage data in finish part",
-                    )
-                    return undefined
-                }
-                // Total input = non-cached + cached (these are separate counts)
-                // Note: cacheWriteInputTokens is not available on finish part
-                const totalInputTokens =
-                    (usage.inputTokens ?? 0) +
-                    (usage.inputTokenDetails?.cacheReadTokens ?? 0)
+                // AI SDK 6 provides totalTokens directly
                 return {
-                    inputTokens: totalInputTokens,
-                    outputTokens: usage.outputTokens ?? 0,
+                    totalTokens: usage?.totalTokens ?? 0,
                     finishReason: (part as any).finishReason,
                 }
             }

app/api/log-feedback/route.ts

@@ -27,9 +27,18 @@ export async function POST(req: Request) {
 
     const { messageId, feedback, sessionId } = data
 
-    // Get user IP for tracking
+    // Skip logging if no sessionId - prevents attaching to wrong user's trace
+    if (!sessionId) {
+        return Response.json({ success: true, logged: false })
+    }
+
+    // Get user IP for tracking (hashed for privacy)
     const forwardedFor = req.headers.get("x-forwarded-for")
-    const userId = forwardedFor?.split(",")[0]?.trim() || "anonymous"
+    const rawIp = forwardedFor?.split(",")[0]?.trim() || "anonymous"
+    const userId =
+        rawIp === "anonymous"
+            ? rawIp
+            : `user-${Buffer.from(rawIp).toString("base64url").slice(0, 8)}`
 
     try {
         // Find the most recent chat trace for this session to attach the score to

app/api/log-save/route.ts

@@ -27,6 +27,11 @@ export async function POST(req: Request) {
 
     const { filename, format, sessionId } = data
 
+    // Skip logging if no sessionId - prevents attaching to wrong user's trace
+    if (!sessionId) {
+        return Response.json({ success: true, logged: false })
+    }
+
     try {
         const timestamp = new Date().toISOString()
 

components/chat-message-display.tsx

@@ -31,6 +31,7 @@ import { getApiEndpoint } from "@/lib/base-path"
 import {
     applyDiagramOperations,
     convertToLegalXml,
+    extractCompleteMxCells,
     isMxCellXmlComplete,
     replaceNodes,
     validateAndFixXml,
@@ -315,12 +316,28 @@ export function ChatMessageDisplay({
 
     const handleDisplayChart = useCallback(
         (xml: string, showToast = false) => {
-            const currentXml = xml || ""
+            let currentXml = xml || ""
+            const startTime = performance.now()
+
+            // During streaming (showToast=false), extract only complete mxCell elements
+            // This allows progressive rendering even with partial/incomplete trailing XML
+            if (!showToast) {
+                const completeCells = extractCompleteMxCells(currentXml)
+                if (!completeCells) {
+                    return
+                }
+                currentXml = completeCells
+            }
+
             const convertedXml = convertToLegalXml(currentXml)
             if (convertedXml !== previousXML.current) {
                 // Parse and validate XML BEFORE calling replaceNodes
                 const parser = new DOMParser()
-                const testDoc = parser.parseFromString(convertedXml, "text/xml")
+                // Wrap in root element for parsing multiple mxCell elements
+                const testDoc = parser.parseFromString(
+                    `<root>${convertedXml}</root>`,
+                    "text/xml",
+                )
                 const parseError = testDoc.querySelector("parsererror")
 
                 if (parseError) {
@@ -347,7 +364,22 @@ export function ChatMessageDisplay({
                         `<mxfile><diagram name="Page-1" id="page-1"><mxGraphModel><root><mxCell id="0"/><mxCell id="1" parent="0"/></root></mxGraphModel></diagram></mxfile>`
                     const replacedXML = replaceNodes(baseXML, convertedXml)
 
-                    // Validate and auto-fix the XML
+                    const xmlProcessTime = performance.now() - startTime
+
+                    // During streaming (showToast=false), skip heavy validation for lower latency
+                    // The quick DOM parse check above catches malformed XML
+                    // Full validation runs on final output (showToast=true)
+                    if (!showToast) {
+                        previousXML.current = convertedXml
+                        const loadStartTime = performance.now()
+                        onDisplayChart(replacedXML, true)
+                        console.log(
+                            `[Streaming] XML processing: ${xmlProcessTime.toFixed(1)}ms, drawio load: ${(performance.now() - loadStartTime).toFixed(1)}ms`,
+                        )
+                        return
+                    }
+
+                    // Final output: run full validation and auto-fix
                     const validation = validateAndFixXml(replacedXML)
                     if (validation.valid) {
                         previousXML.current = convertedXml
@@ -360,18 +392,19 @@ export function ChatMessageDisplay({
                             )
                         }
                         // Skip validation in loadDiagram since we already validated above
+                        const loadStartTime = performance.now()
                         onDisplayChart(xmlToLoad, true)
+                        console.log(
+                            `[Final] XML processing: ${xmlProcessTime.toFixed(1)}ms, validation+load: ${(performance.now() - loadStartTime).toFixed(1)}ms`,
+                        )
                     } else {
                         console.error(
                             "[ChatMessageDisplay] XML validation failed:",
                             validation.error,
                         )
-                        // Only show toast if this is the final XML (not during streaming)
-                        if (showToast) {
-                            toast.error(
-                                "Diagram validation failed. Please try regenerating.",
-                            )
-                        }
+                        toast.error(
+                            "Diagram validation failed. Please try regenerating.",
+                        )
                     }
                 } catch (error) {
                     console.error(
@@ -603,17 +636,10 @@ export function ChatMessageDisplay({
             }
         })
 
-        // Cleanup: clear any pending debounce timeout on unmount
-        return () => {
-            if (debounceTimeoutRef.current) {
-                clearTimeout(debounceTimeoutRef.current)
-                debounceTimeoutRef.current = null
-            }
-            if (editDebounceTimeoutRef.current) {
-                clearTimeout(editDebounceTimeoutRef.current)
-                editDebounceTimeoutRef.current = null
-            }
-        }
+        // NOTE: Don't cleanup debounce timeouts here!
+        // The cleanup runs on every re-render (when messages changes),
+        // which would cancel the timeout before it fires.
+        // Let the timeouts complete naturally - they're harmless if component unmounts.
     }, [messages, handleDisplayChart, chartXML])
 
     const renderToolPart = (part: ToolPartLike) => {

components/chat-panel.tsx

@@ -556,6 +556,23 @@ Continue from EXACTLY where you stopped.`,
             }
         },
         onError: (error) => {
+            // Handle server-side quota limit (429 response)
+            if (error.message.includes("Daily request limit")) {
+                quotaManager.showQuotaLimitToast()
+                return
+            }
+            if (error.message.includes("Daily token limit")) {
+                quotaManager.showTokenLimitToast(dailyTokenLimit)
+                return
+            }
+            if (
+                error.message.includes("Rate limit exceeded") ||
+                error.message.includes("tokens per minute")
+            ) {
+                quotaManager.showTPMLimitToast()
+                return
+            }
+
             // Silence access code error in console since it's handled by UI
             if (!error.message.includes("Invalid or missing access code")) {
                 console.error("Chat error:", error)
@@ -632,22 +649,6 @@ Continue from EXACTLY where you stopped.`,
 
             // DEBUG: Log finish reason to diagnose truncation
             console.log("[onFinish] finishReason:", metadata?.finishReason)
-            console.log("[onFinish] metadata:", metadata)
-
-            if (metadata) {
-                // Use Number.isFinite to guard against NaN (typeof NaN === 'number' is true)
-                const inputTokens = Number.isFinite(metadata.inputTokens)
-                    ? (metadata.inputTokens as number)
-                    : 0
-                const outputTokens = Number.isFinite(metadata.outputTokens)
-                    ? (metadata.outputTokens as number)
-                    : 0
-                const actualTokens = inputTokens + outputTokens
-                if (actualTokens > 0) {
-                    quotaManager.incrementTokenCount(actualTokens)
-                    quotaManager.incrementTPMCount(actualTokens)
-                }
-            }
         },
         sendAutomaticallyWhen: ({ messages }) => {
             const isInContinuationMode = partialXmlRef.current.length > 0
@@ -692,25 +693,6 @@ Continue from EXACTLY where you stopped.`,
                 autoRetryCountRef.current++
             }
 
-            // Check quota limits before auto-retry
-            const tokenLimitCheck = quotaManager.checkTokenLimit()
-            if (!tokenLimitCheck.allowed) {
-                quotaManager.showTokenLimitToast(tokenLimitCheck.used)
-                autoRetryCountRef.current = 0
-                continuationRetryCountRef.current = 0
-                partialXmlRef.current = ""
-                return false
-            }
-
-            const tpmCheck = quotaManager.checkTPMLimit()
-            if (!tpmCheck.allowed) {
-                quotaManager.showTPMLimitToast()
-                autoRetryCountRef.current = 0
-                continuationRetryCountRef.current = 0
-                partialXmlRef.current = ""
-                return false
-            }
-
             return true
         },
     })
@@ -927,9 +909,6 @@ Continue from EXACTLY where you stopped.`,
                 xmlSnapshotsRef.current.set(messageIndex, chartXml)
                 saveXmlSnapshots()
 
-                // Check all quota limits
-                if (!checkAllQuotaLimits()) return
-
                 sendChatMessage(parts, chartXml, previousXml, sessionId)
 
                 // Token count is tracked in onFinish with actual server usage
@@ -1007,30 +986,7 @@ Continue from EXACTLY where you stopped.`,
         saveXmlSnapshots()
     }
 
-    // Check all quota limits (daily requests, tokens, TPM)
-    const checkAllQuotaLimits = (): boolean => {
-        const limitCheck = quotaManager.checkDailyLimit()
-        if (!limitCheck.allowed) {
-            quotaManager.showQuotaLimitToast()
-            return false
-        }
-
-        const tokenLimitCheck = quotaManager.checkTokenLimit()
-        if (!tokenLimitCheck.allowed) {
-            quotaManager.showTokenLimitToast(tokenLimitCheck.used)
-            return false
-        }
-
-        const tpmCheck = quotaManager.checkTPMLimit()
-        if (!tpmCheck.allowed) {
-            quotaManager.showTPMLimitToast()
-            return false
-        }
-
-        return true
-    }
-
-    // Send chat message with headers and increment quota
+    // Send chat message with headers
     const sendChatMessage = (
         parts: any,
         xml: string,
@@ -1080,7 +1036,6 @@ Continue from EXACTLY where you stopped.`,
                 },
             },
         )
-        quotaManager.incrementRequestCount()
     }
 
     // Process files and append content to user text (handles PDF, text, and optionally images)
@@ -1168,13 +1123,8 @@ Continue from EXACTLY where you stopped.`,
             setMessages(newMessages)
         })
 
-        // Check all quota limits
-        if (!checkAllQuotaLimits()) return
-
         // Now send the message after state is guaranteed to be updated
         sendChatMessage(userParts, savedXml, previousXml, sessionId)
-
-        // Token count is tracked in onFinish with actual server usage
     }
 
     const handleEditMessage = async (messageIndex: number, newText: string) => {
@@ -1216,12 +1166,8 @@ Continue from EXACTLY where you stopped.`,
             setMessages(newMessages)
         })
 
-        // Check all quota limits
-        if (!checkAllQuotaLimits()) return
-
         // Now send the edited message after state is guaranteed to be updated
         sendChatMessage(newParts, savedXml, previousXml, sessionId)
-        // Token count is tracked in onFinish with actual server usage
     }
 
     // Collapsed view (desktop only)

lib/dynamo-quota-manager.ts

@@ -0,0 +1,238 @@
+import {
+    ConditionalCheckFailedException,
+    DynamoDBClient,
+    GetItemCommand,
+    UpdateItemCommand,
+} from "@aws-sdk/client-dynamodb"
+
+// Quota tracking is OPT-IN: only enabled if DYNAMODB_QUOTA_TABLE is explicitly set
+// OSS users who don't need quota tracking can simply not set this env var
+const TABLE = process.env.DYNAMODB_QUOTA_TABLE
+const DYNAMODB_REGION = process.env.DYNAMODB_REGION || "ap-northeast-1"
+
+// Only create client if quota is enabled
+const client = TABLE ? new DynamoDBClient({ region: DYNAMODB_REGION }) : null
+
+/**
+ * Check if server-side quota tracking is enabled.
+ * Quota is opt-in: only enabled when DYNAMODB_QUOTA_TABLE env var is set.
+ */
+export function isQuotaEnabled(): boolean {
+    return !!TABLE
+}
+
+interface QuotaLimits {
+    requests: number // Daily request limit
+    tokens: number // Daily token limit
+    tpm: number // Tokens per minute
+}
+
+interface QuotaCheckResult {
+    allowed: boolean
+    error?: string
+    type?: "request" | "token" | "tpm"
+    used?: number
+    limit?: number
+}
+
+/**
+ * Check all quotas and increment request count atomically.
+ * Uses ConditionExpression to prevent race conditions.
+ * Returns which limit was exceeded if any.
+ */
+export async function checkAndIncrementRequest(
+    ip: string,
+    limits: QuotaLimits,
+): Promise<QuotaCheckResult> {
+    // Skip if quota tracking not enabled
+    if (!client || !TABLE) {
+        return { allowed: true }
+    }
+
+    const today = new Date().toISOString().split("T")[0]
+    const currentMinute = Math.floor(Date.now() / 60000).toString()
+    const ttl = Math.floor(Date.now() / 1000) + 7 * 24 * 60 * 60
+
+    try {
+        // Atomic check-and-increment with ConditionExpression
+        // This prevents race conditions by failing if limits are exceeded
+        await client.send(
+            new UpdateItemCommand({
+                TableName: TABLE,
+                Key: { PK: { S: `IP#${ip}` } },
+                // Reset counts if new day/minute, then increment request count
+                UpdateExpression: `
+                    SET lastResetDate = :today,
+                        dailyReqCount = if_not_exists(dailyReqCount, :zero) + :one,
+                        dailyTokenCount = if_not_exists(dailyTokenCount, :zero),
+                        lastMinute = :minute,
+                        tpmCount = if_not_exists(tpmCount, :zero),
+                        #ttl = :ttl
+                `,
+                // Atomic condition: only succeed if ALL limits pass
+                // Uses attribute_not_exists for new items, then checks limits for existing items
+                ConditionExpression: `
+                    (attribute_not_exists(lastResetDate) OR lastResetDate < :today OR
+                     ((attribute_not_exists(dailyReqCount) OR dailyReqCount < :reqLimit) AND
+                      (attribute_not_exists(dailyTokenCount) OR dailyTokenCount < :tokenLimit))) AND
+                    (attribute_not_exists(lastMinute) OR lastMinute <> :minute OR
+                     attribute_not_exists(tpmCount) OR tpmCount < :tpmLimit)
+                `,
+                ExpressionAttributeNames: { "#ttl": "ttl" },
+                ExpressionAttributeValues: {
+                    ":today": { S: today },
+                    ":zero": { N: "0" },
+                    ":one": { N: "1" },
+                    ":minute": { S: currentMinute },
+                    ":ttl": { N: String(ttl) },
+                    ":reqLimit": { N: String(limits.requests || 999999) },
+                    ":tokenLimit": { N: String(limits.tokens || 999999) },
+                    ":tpmLimit": { N: String(limits.tpm || 999999) },
+                },
+            }),
+        )
+
+        return { allowed: true }
+    } catch (e: any) {
+        // Condition failed - need to determine which limit was exceeded
+        if (e instanceof ConditionalCheckFailedException) {
+            // Get current counts to determine which limit was hit
+            try {
+                const getResult = await client.send(
+                    new GetItemCommand({
+                        TableName: TABLE,
+                        Key: { PK: { S: `IP#${ip}` } },
+                    }),
+                )
+
+                const item = getResult.Item
+                const storedDate = item?.lastResetDate?.S
+                const storedMinute = item?.lastMinute?.S
+                const isNewDay = !storedDate || storedDate < today
+
+                const dailyReqCount = isNewDay
+                    ? 0
+                    : Number(item?.dailyReqCount?.N || 0)
+                const dailyTokenCount = isNewDay
+                    ? 0
+                    : Number(item?.dailyTokenCount?.N || 0)
+                const tpmCount =
+                    storedMinute !== currentMinute
+                        ? 0
+                        : Number(item?.tpmCount?.N || 0)
+
+                // Determine which limit was exceeded
+                if (limits.requests > 0 && dailyReqCount >= limits.requests) {
+                    return {
+                        allowed: false,
+                        type: "request",
+                        error: "Daily request limit exceeded",
+                        used: dailyReqCount,
+                        limit: limits.requests,
+                    }
+                }
+                if (limits.tokens > 0 && dailyTokenCount >= limits.tokens) {
+                    return {
+                        allowed: false,
+                        type: "token",
+                        error: "Daily token limit exceeded",
+                        used: dailyTokenCount,
+                        limit: limits.tokens,
+                    }
+                }
+                if (limits.tpm > 0 && tpmCount >= limits.tpm) {
+                    return {
+                        allowed: false,
+                        type: "tpm",
+                        error: "Rate limit exceeded (tokens per minute)",
+                        used: tpmCount,
+                        limit: limits.tpm,
+                    }
+                }
+
+                // Condition failed but no limit clearly exceeded - race condition edge case
+                // Fail safe by allowing (could be a reset race)
+                console.warn(
+                    `[quota] Condition failed but no limit exceeded for IP prefix: ${ip.slice(0, 8)}...`,
+                )
+                return { allowed: true }
+            } catch (getError: any) {
+                console.error(
+                    `[quota] Failed to get quota details after condition failure, IP prefix: ${ip.slice(0, 8)}..., error: ${getError.message}`,
+                )
+                return { allowed: true } // Fail open
+            }
+        }
+
+        // Other DynamoDB errors - fail open
+        console.error(
+            `[quota] DynamoDB error (fail-open), IP prefix: ${ip.slice(0, 8)}..., error: ${e.message}`,
+        )
+        return { allowed: true }
+    }
+}
+
+/**
+ * Record token usage after response completes.
+ * Uses atomic operations to update both daily token count and TPM count.
+ * Handles minute boundaries atomically to prevent race conditions.
+ */
+export async function recordTokenUsage(
+    ip: string,
+    tokens: number,
+): Promise<void> {
+    // Skip if quota tracking not enabled
+    if (!client || !TABLE) return
+    if (!Number.isFinite(tokens) || tokens <= 0) return
+
+    const currentMinute = Math.floor(Date.now() / 60000).toString()
+    const ttl = Math.floor(Date.now() / 1000) + 7 * 24 * 60 * 60
+
+    try {
+        // Try to update assuming same minute (most common case)
+        // Uses condition to ensure we're in the same minute
+        await client.send(
+            new UpdateItemCommand({
+                TableName: TABLE,
+                Key: { PK: { S: `IP#${ip}` } },
+                UpdateExpression:
+                    "SET #ttl = :ttl ADD dailyTokenCount :tokens, tpmCount :tokens",
+                ConditionExpression: "lastMinute = :minute",
+                ExpressionAttributeNames: { "#ttl": "ttl" },
+                ExpressionAttributeValues: {
+                    ":minute": { S: currentMinute },
+                    ":tokens": { N: String(tokens) },
+                    ":ttl": { N: String(ttl) },
+                },
+            }),
+        )
+    } catch (e: any) {
+        if (e instanceof ConditionalCheckFailedException) {
+            // Different minute - reset TPM count and set new minute
+            try {
+                await client.send(
+                    new UpdateItemCommand({
+                        TableName: TABLE,
+                        Key: { PK: { S: `IP#${ip}` } },
+                        UpdateExpression:
+                            "SET lastMinute = :minute, tpmCount = :tokens, #ttl = :ttl ADD dailyTokenCount :tokens",
+                        ExpressionAttributeNames: { "#ttl": "ttl" },
+                        ExpressionAttributeValues: {
+                            ":minute": { S: currentMinute },
+                            ":tokens": { N: String(tokens) },
+                            ":ttl": { N: String(ttl) },
+                        },
+                    }),
+                )
+            } catch (retryError: any) {
+                console.error(
+                    `[quota] Failed to record tokens (retry), IP prefix: ${ip.slice(0, 8)}..., tokens: ${tokens}, error: ${retryError.message}`,
+                )
+            }
+        } else {
+            console.error(
+                `[quota] Failed to record tokens, IP prefix: ${ip.slice(0, 8)}..., tokens: ${tokens}, error: ${e.message}`,
+            )
+        }
+    }
+}

lib/langfuse.ts

@@ -21,9 +21,11 @@ export function getLangfuseClient(): LangfuseClient | null {
     return langfuseClient
 }
 
-// Check if Langfuse is configured
+// Check if Langfuse is configured (both keys required)
 export function isLangfuseEnabled(): boolean {
-    return !!process.env.LANGFUSE_PUBLIC_KEY
+    return !!(
+        process.env.LANGFUSE_PUBLIC_KEY && process.env.LANGFUSE_SECRET_KEY
+    )
 }
 
 // Update trace with input data at the start of request
@@ -43,34 +45,16 @@ export function setTraceInput(params: {
 }
 
 // Update trace with output and end the span
-export function setTraceOutput(
-    output: string,
-    usage?: { promptTokens?: number; completionTokens?: number },
-) {
+// Note: AI SDK 6 telemetry automatically reports token usage on its spans,
+// so we only need to set the output text and close our wrapper span
+export function setTraceOutput(output: string) {
     if (!isLangfuseEnabled()) return
 
     updateActiveTrace({ output })
 
+    // End the observe() wrapper span (AI SDK creates its own child spans with usage)
     const activeSpan = api.trace.getActiveSpan()
     if (activeSpan) {
-        // Manually set usage attributes since AI SDK Bedrock streaming doesn't provide them
-        if (usage?.promptTokens) {
-            activeSpan.setAttribute("ai.usage.promptTokens", usage.promptTokens)
-            activeSpan.setAttribute(
-                "gen_ai.usage.input_tokens",
-                usage.promptTokens,
-            )
-        }
-        if (usage?.completionTokens) {
-            activeSpan.setAttribute(
-                "ai.usage.completionTokens",
-                usage.completionTokens,
-            )
-            activeSpan.setAttribute(
-                "gen_ai.usage.output_tokens",
-                usage.completionTokens,
-            )
-        }
         activeSpan.end()
     }
 }

lib/use-quota-manager.tsx

@@ -1,11 +1,10 @@
 "use client"
 
-import { useCallback, useMemo } from "react"
+import { useCallback } from "react"
 import { toast } from "sonner"
 import { QuotaLimitToast } from "@/components/quota-limit-toast"
 import { useDictionary } from "@/hooks/use-dictionary"
 import { formatMessage } from "@/lib/i18n/utils"
-import { STORAGE_KEYS } from "@/lib/storage"
 
 export interface QuotaConfig {
     dailyRequestLimit: number
@@ -13,134 +12,19 @@ export interface QuotaConfig {
     tpmLimit: number
 }
 
-export interface QuotaCheckResult {
-    allowed: boolean
-    remaining: number
-    used: number
-}
-
 /**
- * Hook for managing request/token quotas and rate limiting.
- * Handles three types of limits:
- * - Daily request limit
- * - Daily token limit
- * - Tokens per minute (TPM) rate limit
- *
- * Users with their own API key bypass all limits.
+ * Hook for displaying quota limit toasts.
+ * Server-side handles actual quota enforcement via DynamoDB.
+ * This hook only provides UI feedback when limits are exceeded.
  */
 export function useQuotaManager(config: QuotaConfig): {
-    hasOwnApiKey: () => boolean
-    checkDailyLimit: () => QuotaCheckResult
-    checkTokenLimit: () => QuotaCheckResult
-    checkTPMLimit: () => QuotaCheckResult
-    incrementRequestCount: () => void
-    incrementTokenCount: (tokens: number) => void
-    incrementTPMCount: (tokens: number) => void
     showQuotaLimitToast: () => void
     showTokenLimitToast: (used: number) => void
     showTPMLimitToast: () => void
 } {
     const { dailyRequestLimit, dailyTokenLimit, tpmLimit } = config
-
     const dict = useDictionary()
 
-    // Check if user has their own API key configured (bypass limits)
-    const hasOwnApiKey = useCallback((): boolean => {
-        const provider = localStorage.getItem(STORAGE_KEYS.aiProvider)
-        const apiKey = localStorage.getItem(STORAGE_KEYS.aiApiKey)
-        return !!(provider && apiKey)
-    }, [])
-
-    // Generic helper: Parse count from localStorage with NaN guard
-    const parseStorageCount = (key: string): number => {
-        const count = parseInt(localStorage.getItem(key) || "0", 10)
-        return Number.isNaN(count) ? 0 : count
-    }
-
-    // Generic helper: Create quota checker factory
-    const createQuotaChecker = useCallback(
-        (
-            getTimeKey: () => string,
-            timeStorageKey: string,
-            countStorageKey: string,
-            limit: number,
-        ) => {
-            return (): QuotaCheckResult => {
-                if (hasOwnApiKey())
-                    return { allowed: true, remaining: -1, used: 0 }
-                if (limit <= 0) return { allowed: true, remaining: -1, used: 0 }
-
-                const currentTime = getTimeKey()
-                const storedTime = localStorage.getItem(timeStorageKey)
-                let count = parseStorageCount(countStorageKey)
-
-                if (storedTime !== currentTime) {
-                    count = 0
-                    localStorage.setItem(timeStorageKey, currentTime)
-                    localStorage.setItem(countStorageKey, "0")
-                }
-
-                return {
-                    allowed: count < limit,
-                    remaining: limit - count,
-                    used: count,
-                }
-            }
-        },
-        [hasOwnApiKey],
-    )
-
-    // Generic helper: Create quota incrementer factory
-    const createQuotaIncrementer = useCallback(
-        (
-            getTimeKey: () => string,
-            timeStorageKey: string,
-            countStorageKey: string,
-            validateInput: boolean = false,
-        ) => {
-            return (tokens: number = 1): void => {
-                if (validateInput && (!Number.isFinite(tokens) || tokens <= 0))
-                    return
-
-                const currentTime = getTimeKey()
-                const storedTime = localStorage.getItem(timeStorageKey)
-                let count = parseStorageCount(countStorageKey)
-
-                if (storedTime !== currentTime) {
-                    count = 0
-                    localStorage.setItem(timeStorageKey, currentTime)
-                }
-
-                localStorage.setItem(countStorageKey, String(count + tokens))
-            }
-        },
-        [],
-    )
-
-    // Check daily request limit
-    const checkDailyLimit = useMemo(
-        () =>
-            createQuotaChecker(
-                () => new Date().toDateString(),
-                STORAGE_KEYS.requestDate,
-                STORAGE_KEYS.requestCount,
-                dailyRequestLimit,
-            ),
-        [createQuotaChecker, dailyRequestLimit],
-    )
-
-    // Increment request count
-    const incrementRequestCount = useMemo(
-        () =>
-            createQuotaIncrementer(
-                () => new Date().toDateString(),
-                STORAGE_KEYS.requestDate,
-                STORAGE_KEYS.requestCount,
-                false,
-            ),
-        [createQuotaIncrementer],
-    )
-
     // Show quota limit toast (request-based)
     const showQuotaLimitToast = useCallback(() => {
         toast.custom(
@@ -155,30 +39,6 @@ export function useQuotaManager(config: QuotaConfig): {
         )
     }, [dailyRequestLimit])
 
-    // Check daily token limit
-    const checkTokenLimit = useMemo(
-        () =>
-            createQuotaChecker(
-                () => new Date().toDateString(),
-                STORAGE_KEYS.tokenDate,
-                STORAGE_KEYS.tokenCount,
-                dailyTokenLimit,
-            ),
-        [createQuotaChecker, dailyTokenLimit],
-    )
-
-    // Increment token count
-    const incrementTokenCount = useMemo(
-        () =>
-            createQuotaIncrementer(
-                () => new Date().toDateString(),
-                STORAGE_KEYS.tokenDate,
-                STORAGE_KEYS.tokenCount,
-                true, // Validate input tokens
-            ),
-        [createQuotaIncrementer],
-    )
-
     // Show token limit toast
     const showTokenLimitToast = useCallback(
         (used: number) => {
@@ -197,30 +57,6 @@ export function useQuotaManager(config: QuotaConfig): {
         [dailyTokenLimit],
     )
 
-    // Check TPM (tokens per minute) limit
-    const checkTPMLimit = useMemo(
-        () =>
-            createQuotaChecker(
-                () => Math.floor(Date.now() / 60000).toString(),
-                STORAGE_KEYS.tpmMinute,
-                STORAGE_KEYS.tpmCount,
-                tpmLimit,
-            ),
-        [createQuotaChecker, tpmLimit],
-    )
-
-    // Increment TPM count
-    const incrementTPMCount = useMemo(
-        () =>
-            createQuotaIncrementer(
-                () => Math.floor(Date.now() / 60000).toString(),
-                STORAGE_KEYS.tpmMinute,
-                STORAGE_KEYS.tpmCount,
-                true, // Validate input tokens
-            ),
-        [createQuotaIncrementer],
-    )
-
     // Show TPM limit toast
     const showTPMLimitToast = useCallback(() => {
         const limitDisplay =
@@ -233,18 +69,6 @@ export function useQuotaManager(config: QuotaConfig): {
     }, [tpmLimit, dict])
 
     return {
-        // Check functions
-        hasOwnApiKey,
-        checkDailyLimit,
-        checkTokenLimit,
-        checkTPMLimit,
-
-        // Increment functions
-        incrementRequestCount,
-        incrementTokenCount,
-        incrementTPMCount,
-
-        // Toast functions
         showQuotaLimitToast,
         showTokenLimitToast,
         showTPMLimitToast,

lib/utils.ts

@@ -61,6 +61,47 @@ export function isMxCellXmlComplete(xml: string | undefined | null): boolean {
     return trimmed.endsWith("/>") || trimmed.endsWith("</mxCell>")
 }
 
+/**
+ * Extract only complete mxCell elements from partial/streaming XML.
+ * This allows progressive rendering during streaming by ignoring incomplete trailing elements.
+ * @param xml - The partial XML string (may contain incomplete trailing mxCell)
+ * @returns XML string containing only complete mxCell elements
+ */
+export function extractCompleteMxCells(xml: string | undefined | null): string {
+    if (!xml) return ""
+
+    const completeCells: Array<{ index: number; text: string }> = []
+
+    // Match self-closing mxCell tags: <mxCell ... />
+    // Also match mxCell with nested mxGeometry: <mxCell ...>...<mxGeometry .../></mxCell>
+    const selfClosingPattern = /<mxCell\s+[^>]*\/>/g
+    const nestedPattern = /<mxCell\s+[^>]*>[\s\S]*?<\/mxCell>/g
+
+    // Find all self-closing mxCell elements
+    let match: RegExpExecArray | null
+    while ((match = selfClosingPattern.exec(xml)) !== null) {
+        completeCells.push({ index: match.index, text: match[0] })
+    }
+
+    // Find all mxCell elements with nested content (like mxGeometry)
+    while ((match = nestedPattern.exec(xml)) !== null) {
+        completeCells.push({ index: match.index, text: match[0] })
+    }
+
+    // Sort by position to maintain order
+    completeCells.sort((a, b) => a.index - b.index)
+
+    // Remove duplicates (a self-closing match might overlap with nested match)
+    const seen = new Set<number>()
+    const uniqueCells = completeCells.filter((cell) => {
+        if (seen.has(cell.index)) return false
+        seen.add(cell.index)
+        return true
+    })
+
+    return uniqueCells.map((c) => c.text).join("\n")
+}
+
 // ============================================================================
 // XML Parsing Helpers
 // ============================================================================

package.json

@@ -32,6 +32,7 @@
         "@ai-sdk/google": "^3.0.0",
         "@ai-sdk/openai": "^3.0.0",
         "@ai-sdk/react": "^3.0.1",
+        "@aws-sdk/client-dynamodb": "^3.957.0",
         "@aws-sdk/credential-providers": "^3.943.0",
         "@formatjs/intl-localematcher": "^0.7.2",
         "@langfuse/client": "^4.4.9",