Merge pull request #1206 from mykonos-ibiza/limit-running-agents

feat(agent-run-limits): implement agent run limit checks and UI dialogs

backend/agent/api.py

@@ -25,6 +25,7 @@ from utils.constants import MODEL_NAME_ALIASES
 from flags.flags import is_enabled
 
 from .config_helper import extract_agent_config, build_unified_config, extract_tools_for_agent_run, get_mcp_configs
+from .utils import check_agent_run_limit
 from .versioning.version_service import get_version_service
 from .versioning.api import router as version_router, initialize as initialize_versioning
 
@@ -437,6 +438,18 @@ async def start_agent(
     if not can_run:
         raise HTTPException(status_code=402, detail={"message": message, "subscription": subscription})
 
+    # Check agent run limit (maximum parallel runs in past 24 hours)
+    limit_check = await check_agent_run_limit(client, account_id)
+    if not limit_check['can_start']:
+        error_detail = {
+            "message": f"Maximum of {config.MAX_PARALLEL_AGENT_RUNS} parallel agent runs allowed within 24 hours. You currently have {limit_check['running_count']} running.",
+            "running_thread_ids": limit_check['running_thread_ids'],
+            "running_count": limit_check['running_count'],
+            "limit": config.MAX_PARALLEL_AGENT_RUNS
+        }
+        logger.warning(f"Agent run limit exceeded for account {account_id}: {limit_check['running_count']} running agents")
+        raise HTTPException(status_code=429, detail=error_detail)
+
     try:
         project_result = await client.table('projects').select('*').eq('project_id', project_id).execute()
         if not project_result.data:
@@ -1049,6 +1062,18 @@ async def initiate_agent_with_files(
     if not can_run:
         raise HTTPException(status_code=402, detail={"message": message, "subscription": subscription})
 
+    # Check agent run limit (maximum parallel runs in past 24 hours)
+    limit_check = await check_agent_run_limit(client, account_id)
+    if not limit_check['can_start']:
+        error_detail = {
+            "message": f"Maximum of {config.MAX_PARALLEL_AGENT_RUNS} parallel agent runs allowed within 24 hours. You currently have {limit_check['running_count']} running.",
+            "running_thread_ids": limit_check['running_thread_ids'],
+            "running_count": limit_check['running_count'],
+            "limit": config.MAX_PARALLEL_AGENT_RUNS
+        }
+        logger.warning(f"Agent run limit exceeded for account {account_id}: {limit_check['running_count']} running agents")
+        raise HTTPException(status_code=429, detail=error_detail)
+
     try:
         # 1. Create Project
         placeholder_name = f"{prompt[:30]}..." if len(prompt) > 30 else prompt

backend/agent/utils.py

@@ -1,7 +1,10 @@
 import json
-from typing import Optional
+from typing import Optional, List, Dict, Any
+from datetime import datetime, timezone, timedelta
 from utils.logger import logger
+from utils.config import config
 from services import redis
+from run_agent_background import update_agent_run_status
 
 
 async def _cleanup_redis_response_list(agent_run_id: str):
@@ -67,11 +70,64 @@ async def stop_agent_run(db, agent_run_id: str, error_message: Optional[str] = N
                 except Exception as e:
                     logger.warning(f"Failed to publish STOP signal to instance channel {instance_control_channel}: {str(e)}")
             else:
-                 logger.warning(f"Unexpected key format found: {key}")
+                logger.warning(f"Unexpected key format found: {key}")
 
         await _cleanup_redis_response_list(agent_run_id)
 
     except Exception as e:
         logger.error(f"Failed to find or signal active instances for {agent_run_id}: {str(e)}")
 
-    logger.info(f"Successfully initiated stop process for agent run: {agent_run_id}") 
+    logger.info(f"Successfully initiated stop process for agent run: {agent_run_id}")
+
+
+async def check_agent_run_limit(client, account_id: str) -> Dict[str, Any]:
+    """
+    Check if the account has reached the limit of 3 parallel agent runs within the past 24 hours.
+    
+    Returns:
+        Dict with 'can_start' (bool), 'running_count' (int), 'running_thread_ids' (list)
+    """
+    try:
+        # Calculate 24 hours ago
+        twenty_four_hours_ago = datetime.now(timezone.utc) - timedelta(hours=24)
+        twenty_four_hours_ago_iso = twenty_four_hours_ago.isoformat()
+        
+        logger.debug(f"Checking agent run limit for account {account_id} since {twenty_four_hours_ago_iso}")
+        
+        # Get all threads for this account
+        threads_result = await client.table('threads').select('thread_id').eq('account_id', account_id).execute()
+        
+        if not threads_result.data:
+            logger.debug(f"No threads found for account {account_id}")
+            return {
+                'can_start': True,
+                'running_count': 0,
+                'running_thread_ids': []
+            }
+        
+        thread_ids = [thread['thread_id'] for thread in threads_result.data]
+        logger.debug(f"Found {len(thread_ids)} threads for account {account_id}")
+        
+        # Query for running agent runs within the past 24 hours for these threads
+        running_runs_result = await client.table('agent_runs').select('id', 'thread_id', 'started_at').in_('thread_id', thread_ids).eq('status', 'running').gte('started_at', twenty_four_hours_ago_iso).execute()
+        
+        running_runs = running_runs_result.data or []
+        running_count = len(running_runs)
+        running_thread_ids = [run['thread_id'] for run in running_runs]
+        
+        logger.info(f"Account {account_id} has {running_count} running agent runs in the past 24 hours")
+        
+        return {
+            'can_start': running_count < config.MAX_PARALLEL_AGENT_RUNS,
+            'running_count': running_count,
+            'running_thread_ids': running_thread_ids
+        }
+        
+    except Exception as e:
+        logger.error(f"Error checking agent run limit for account {account_id}: {str(e)}")
+        # In case of error, allow the run to proceed but log the error
+        return {
+            'can_start': True,
+            'running_count': 0,
+            'running_thread_ids': []
+        } 

backend/utils/config.py

@@ -267,7 +267,34 @@ class Configuration:
     # API Keys system configuration
     API_KEY_SECRET: str = "default-secret-key-change-in-production"
     API_KEY_LAST_USED_THROTTLE_SECONDS: int = 900
+    
+    # Agent execution limits (can be overridden via environment variable)
+    _MAX_PARALLEL_AGENT_RUNS_ENV: Optional[str] = None
 
+    @property
+    def MAX_PARALLEL_AGENT_RUNS(self) -> int:
+        """
+        Get the maximum parallel agent runs limit.
+        
+        Can be overridden via MAX_PARALLEL_AGENT_RUNS environment variable.
+        Defaults:
+        - Production: 3
+        - Local/Staging: 999999 (effectively infinite)
+        """
+        # Check for environment variable override first
+        if self._MAX_PARALLEL_AGENT_RUNS_ENV is not None:
+            try:
+                return int(self._MAX_PARALLEL_AGENT_RUNS_ENV)
+            except ValueError:
+                logger.warning(f"Invalid MAX_PARALLEL_AGENT_RUNS value: {self._MAX_PARALLEL_AGENT_RUNS_ENV}, using default")
+        
+        # Environment-based defaults
+        if self.ENV_MODE == EnvMode.PRODUCTION:
+            return 3
+        else:
+            # Local and staging: effectively infinite
+            return 999999
+    
     @property
     def STRIPE_PRODUCT_ID(self) -> str:
         if self.ENV_MODE == EnvMode.STAGING:
@@ -317,6 +344,11 @@ class Configuration:
                 else:
                     # String or other type
                     setattr(self, key, env_val)
+        
+        # Custom handling for environment-dependent properties
+        max_parallel_runs_env = os.getenv("MAX_PARALLEL_AGENT_RUNS")
+        if max_parallel_runs_env is not None:
+            self._MAX_PARALLEL_AGENT_RUNS_ENV = max_parallel_runs_env
     
     def _validate(self):
         """Validate configuration based on type hints."""

frontend/src/app/(dashboard)/projects/[projectId]/thread/[threadId]/page.tsx

@@ -8,7 +8,7 @@ import React, {
   useMemo,
 } from 'react';
 import { useSearchParams } from 'next/navigation';
-import { BillingError } from '@/lib/api';
+import { BillingError, AgentRunLimitError } from '@/lib/api';
 import { toast } from 'sonner';
 import { ChatInput } from '@/components/thread/chat-input/chat-input';
 import { useSidebar } from '@/components/ui/sidebar';
@@ -28,6 +28,7 @@ import { useThreadData, useToolCalls, useBilling, useKeyboardShortcuts } from '.
 import { ThreadError, UpgradeDialog, ThreadLayout } from '../_components';
 import { useVncPreloader } from '@/hooks/useVncPreloader';
 import { useThreadAgent } from '@/hooks/react-query/agents/use-agents';
+import { AgentRunLimitDialog } from '@/components/thread/agent-run-limit-dialog';
 
 export default function ThreadPage({
   params,
@@ -55,6 +56,11 @@ export default function ThreadPage({
   const [isSidePanelAnimating, setIsSidePanelAnimating] = useState(false);
   const [userInitiatedRun, setUserInitiatedRun] = useState(false);
   const [showScrollToBottom, setShowScrollToBottom] = useState(false);
+  const [showAgentLimitDialog, setShowAgentLimitDialog] = useState(false);
+  const [agentLimitData, setAgentLimitData] = useState<{
+    runningCount: number;
+    runningThreadIds: string[];
+  } | null>(null);
   
 
   // Refs - simplified for flex-column-reverse
@@ -320,6 +326,21 @@ export default function ThreadPage({
             return;
           }
 
+          if (error instanceof AgentRunLimitError) {
+            console.log("Caught AgentRunLimitError:", error.detail);
+            const { running_thread_ids, running_count } = error.detail;
+            
+            // Show the dialog with limit information
+            setAgentLimitData({
+              runningCount: running_count,
+              runningThreadIds: running_thread_ids,
+            });
+            setShowAgentLimitDialog(true);
+
+            setMessages(prev => prev.filter(m => m.message_id !== optimisticUserMessage.message_id));
+            return;
+          }
+
           throw new Error(`Failed to start agent: ${error?.message || error}`);
         }
 
@@ -331,7 +352,7 @@ export default function ThreadPage({
 
       } catch (err) {
         console.error('Error sending message or starting agent:', err);
-        if (!(err instanceof BillingError)) {
+        if (!(err instanceof BillingError) && !(err instanceof AgentRunLimitError)) {
           toast.error(err instanceof Error ? err.message : 'Operation failed');
         }
         setMessages((prev) =>
@@ -735,6 +756,16 @@ export default function ThreadPage({
         onOpenChange={setShowUpgradeDialog}
         onDismiss={handleDismissUpgradeDialog}
       />
+
+      {agentLimitData && (
+        <AgentRunLimitDialog
+          open={showAgentLimitDialog}
+          onOpenChange={setShowAgentLimitDialog}
+          runningCount={agentLimitData.runningCount}
+          runningThreadIds={agentLimitData.runningThreadIds}
+          projectId={projectId}
+        />
+      )}
     </>
   );
 } 

frontend/src/components/dashboard/dashboard-content.tsx

@@ -10,6 +10,7 @@ import {
 } from '@/components/thread/chat-input/chat-input';
 import {
   BillingError,
+  AgentRunLimitError,
 } from '@/lib/api';
 import { useIsMobile } from '@/hooks/use-mobile';
 import { useSidebar } from '@/components/ui/sidebar';
@@ -32,6 +33,7 @@ import { Examples } from './examples';
 import { useThreadQuery } from '@/hooks/react-query/threads/use-threads';
 import { normalizeFilenameToNFC } from '@/lib/utils/unicode';
 import { KortixLogo } from '../sidebar/kortix-logo';
+import { AgentRunLimitDialog } from '@/components/thread/agent-run-limit-dialog';
 
 const PENDING_PROMPT_KEY = 'pendingAgentPrompt';
 
@@ -43,6 +45,11 @@ export function DashboardContent() {
   const [initiatedThreadId, setInitiatedThreadId] = useState<string | null>(null);
   const { billingError, handleBillingError, clearBillingError } =
     useBillingError();
+  const [showAgentLimitDialog, setShowAgentLimitDialog] = useState(false);
+  const [agentLimitData, setAgentLimitData] = useState<{
+    runningCount: number;
+    runningThreadIds: string[];
+  } | null>(null);
   const router = useRouter();
   const searchParams = useSearchParams();
   const isMobile = useIsMobile();
@@ -153,6 +160,16 @@ export function DashboardContent() {
       if (error instanceof BillingError) {
         console.log('Handling BillingError:', error.detail);
         onOpen("paymentRequiredDialog");
+      } else if (error instanceof AgentRunLimitError) {
+        console.log('Handling AgentRunLimitError:', error.detail);
+        const { running_thread_ids, running_count } = error.detail;
+        
+        // Show the dialog with limit information
+        setAgentLimitData({
+          runningCount: running_count,
+          runningThreadIds: running_thread_ids,
+        });
+        setShowAgentLimitDialog(true);
       }
       setIsSubmitting(false);
     }
@@ -257,6 +274,16 @@ export function DashboardContent() {
           isOpen={!!billingError}
         />
       </div>
+
+      {agentLimitData && (
+        <AgentRunLimitDialog
+          open={showAgentLimitDialog}
+          onOpenChange={setShowAgentLimitDialog}
+          runningCount={agentLimitData.runningCount}
+          runningThreadIds={agentLimitData.runningThreadIds}
+          projectId={undefined} // Dashboard doesn't have a specific project context
+        />
+      )}
     </>
   );
 }

frontend/src/components/home/sections/hero-section.tsx

@@ -11,6 +11,7 @@ import { useRouter } from 'next/navigation';
 import { useAuth } from '@/components/AuthProvider';
 import {
   BillingError,
+  AgentRunLimitError,
 } from '@/lib/api';
 import { useInitiateAgentMutation } from '@/hooks/react-query/dashboard/use-initiate-agent';
 import { useThreadQuery } from '@/hooks/react-query/threads/use-threads';
@@ -37,6 +38,7 @@ import { normalizeFilenameToNFC } from '@/lib/utils/unicode';
 import { createQueryHook } from '@/hooks/use-query';
 import { agentKeys } from '@/hooks/react-query/agents/keys';
 import { getAgents } from '@/hooks/react-query/agents/utils';
+import { AgentRunLimitDialog } from '@/components/thread/agent-run-limit-dialog';
 
 // Custom dialog overlay with blur effect
 const BlurredDialogOverlay = () => (
@@ -67,6 +69,11 @@ export function HeroSection() {
   const [initiatedThreadId, setInitiatedThreadId] = useState<string | null>(null);
   const threadQuery = useThreadQuery(initiatedThreadId || '');
   const chatInputRef = useRef<ChatInputHandles>(null);
+  const [showAgentLimitDialog, setShowAgentLimitDialog] = useState(false);
+  const [agentLimitData, setAgentLimitData] = useState<{
+    runningCount: number;
+    runningThreadIds: string[];
+  } | null>(null);
 
   // Fetch agents for selection
   const { data: agentsResponse } = createQueryHook(
@@ -199,6 +206,16 @@ export function HeroSection() {
       if (error instanceof BillingError) {
         console.log('Billing error:', error.detail);
         onOpen("paymentRequiredDialog");
+      } else if (error instanceof AgentRunLimitError) {
+        console.log('Handling AgentRunLimitError:', error.detail);
+        const { running_thread_ids, running_count } = error.detail;
+        
+        // Show the dialog with limit information
+        setAgentLimitData({
+          runningCount: running_count,
+          runningThreadIds: running_thread_ids,
+        });
+        setShowAgentLimitDialog(true);
       } else {
         const isConnectionError =
           error instanceof TypeError &&
@@ -414,6 +431,16 @@ export function HeroSection() {
         onDismiss={clearBillingError}
         isOpen={!!billingError}
       />
+
+      {agentLimitData && (
+        <AgentRunLimitDialog
+          open={showAgentLimitDialog}
+          onOpenChange={setShowAgentLimitDialog}
+          runningCount={agentLimitData.runningCount}
+          runningThreadIds={agentLimitData.runningThreadIds}
+          projectId={undefined} // Hero section doesn't have a specific project context
+        />
+      )}
     </section>
   );
 }

frontend/src/components/thread/agent-run-limit-dialog.tsx

@@ -0,0 +1,328 @@
+'use client';
+
+import React, { useState, useMemo } from 'react';
+import { AlertTriangle, ExternalLink, X, Square, Loader2 } from 'lucide-react';
+import { Button } from '@/components/ui/button';
+import {
+  Dialog,
+  DialogContent,
+  DialogDescription,
+  DialogHeader,
+  DialogTitle,
+} from '@/components/ui/dialog';
+import { Badge } from '@/components/ui/badge';
+import { Card, CardContent } from '@/components/ui/card';
+import { Separator } from '@/components/ui/separator';
+import { Tooltip, TooltipContent, TooltipTrigger } from '@/components/ui/tooltip';
+import Link from 'next/link';
+import { useStopAgentMutation } from '@/hooks/react-query/threads/use-agent-run';
+import { AgentRun, getAgentRuns } from '@/lib/api';
+import { toast } from 'sonner';
+import { useQueries, useQueryClient } from '@tanstack/react-query';
+import { getThread, getProject } from '@/hooks/react-query/threads/utils';
+import { threadKeys } from '@/hooks/react-query/threads/keys';
+
+interface RunningThreadInfo {
+  threadId: string;
+  name: string;
+  projectId: string | null;
+  projectName: string;
+  agentRun: AgentRun | null;
+  isLoading: boolean;
+  error?: boolean;
+}
+
+interface RunningThreadItemProps {
+  threadInfo: RunningThreadInfo;
+  onThreadStopped: () => void;
+}
+
+const RunningThreadItem: React.FC<RunningThreadItemProps> = ({
+  threadInfo,
+  onThreadStopped,
+}) => {
+  const stopAgentMutation = useStopAgentMutation();
+  const queryClient = useQueryClient();
+
+  const handleStop = async () => {
+    if (!threadInfo.agentRun?.id) return;
+    
+    try {
+      await stopAgentMutation.mutateAsync(threadInfo.agentRun.id);
+      
+      // Invalidate relevant queries to refetch updated data
+      await Promise.all([
+        // Refetch agent runs for this thread to update the running status
+        queryClient.invalidateQueries({
+          queryKey: threadKeys.agentRuns(threadInfo.threadId)
+        }),
+        // Refetch thread details in case the status affects the thread
+        queryClient.invalidateQueries({
+          queryKey: threadKeys.details(threadInfo.threadId)
+        })
+      ]);
+      
+      toast.success('Agent stopped successfully');
+      onThreadStopped();
+    } catch (error) {
+      console.error('Failed to stop agent:', error);
+      toast.error('Failed to stop agent');
+    }
+  };
+
+  const getThreadDisplayName = () => {
+    if (threadInfo.name && threadInfo.name.trim()) {
+      return threadInfo.name;
+    }
+    return `Thread ${threadInfo.threadId.slice(0, 8)}...`;
+  };
+
+  const getProjectDisplayName = () => {
+    if (threadInfo.projectName && threadInfo.projectName.trim()) {
+      return threadInfo.projectName;
+    }
+    return threadInfo.projectId ? `Project ${threadInfo.projectId.slice(0, 8)}...` : 'No Project';
+  };
+
+  return (
+    <div className="flex items-center justify-between rounded-lg border bg-muted/50 px-3 py-2 gap-2">
+      <div className="flex items-center gap-2 min-w-0 flex-1">
+        <div className="h-2 w-2 rounded-full bg-green-500 animate-pulse" />
+        <div className="min-w-0 flex-1">
+          <div className="font-medium text-sm truncate">
+            {getProjectDisplayName()}
+          </div>
+          <code className="text-xs text-muted-foreground/70 truncate block">
+            {threadInfo.threadId}
+          </code>
+        </div>
+      </div>
+      
+      <div className="flex items-center gap-1">
+        {threadInfo.agentRun && (
+          <Tooltip>
+            <TooltipTrigger asChild>
+              <Button
+                variant="ghost"
+                size="sm"
+                className="h-7 w-7 p-0 hover:bg-red-100 hover:text-red-600"
+                onClick={handleStop}
+                disabled={stopAgentMutation.isPending || threadInfo.isLoading}
+              >
+                {stopAgentMutation.isPending ? (
+                  <Loader2 className="h-3 w-3 animate-spin" />
+                ) : (
+                  <Square className="h-3 w-3" />
+                )}
+                <span className="sr-only">Stop agent</span>
+              </Button>
+            </TooltipTrigger>
+            <TooltipContent>Stop this agent</TooltipContent>
+          </Tooltip>
+        )}
+        
+        {threadInfo.projectId && (
+          <Tooltip>
+            <TooltipTrigger asChild>
+              <Button
+                asChild
+                variant="ghost"
+                size="sm"
+                className="h-7 w-7 p-0 hover:bg-background"
+              >
+                <Link href={`/projects/${threadInfo.projectId}/thread/${threadInfo.threadId}`} target="_blank">
+                  <ExternalLink className="h-3 w-3" />
+                  <span className="sr-only">Open thread</span>
+                </Link>
+              </Button>
+            </TooltipTrigger>
+            <TooltipContent>Open thread</TooltipContent>
+          </Tooltip>
+        )}
+      </div>
+    </div>
+  );
+};
+
+interface AgentRunLimitDialogProps {
+  open: boolean;
+  onOpenChange: (open: boolean) => void;
+  runningCount: number;
+  runningThreadIds: string[];
+  projectId?: string;
+}
+
+export const AgentRunLimitDialog: React.FC<AgentRunLimitDialogProps> = ({
+  open,
+  onOpenChange,
+  runningCount,
+  runningThreadIds,
+  projectId,
+}) => {
+  // Use the exact same query keys as existing hooks for perfect cache consistency with sidebar
+  // This means data fetched by the sidebar is immediately available here without additional requests
+  const threadQueries = useQueries({
+    queries: runningThreadIds.map(threadId => ({
+      queryKey: threadKeys.details(threadId),
+      queryFn: () => getThread(threadId),
+      enabled: open && !!threadId,
+      retry: 1,
+    }))
+  });
+
+  const agentRunQueries = useQueries({
+    queries: runningThreadIds.map(threadId => ({
+      queryKey: threadKeys.agentRuns(threadId), // This matches useAgentRunsQuery exactly
+      queryFn: () => getAgentRuns(threadId),
+      enabled: open && !!threadId,
+      retry: 1,
+    }))
+  });
+
+  // Use the same query keys as useProjectQuery for cache consistency
+  const projectQueries = useQueries({
+    queries: runningThreadIds.map(threadId => {
+      const threadQuery = threadQueries.find((_, index) => runningThreadIds[index] === threadId);
+      const projectId = threadQuery?.data?.project_id;
+      
+      return {
+        queryKey: threadKeys.project(projectId || ""),
+        queryFn: () => projectId ? getProject(projectId) : null,
+        enabled: open && !!projectId,
+        retry: 1,
+      };
+    })
+  });
+
+  // Process the React Query results into our thread info structure
+  const runningThreadsInfo: RunningThreadInfo[] = useMemo(() => {
+    return runningThreadIds.map((threadId, index) => {
+      const threadQuery = threadQueries[index];
+      const agentRunQuery = agentRunQueries[index];
+      const projectQuery = projectQueries[index];
+      
+      const isLoading = threadQuery.isLoading || agentRunQuery.isLoading || projectQuery.isLoading;
+      const hasError = threadQuery.isError || agentRunQuery.isError || projectQuery.isError;
+      
+      // Find the running agent run for this thread
+      const runningAgentRun = agentRunQuery.data?.find((run: AgentRun) => run.status === 'running') || null;
+      
+      // Get thread name from first user message if available
+      let threadName = '';
+      if (threadQuery.data?.messages?.length > 0) {
+        const firstUserMessage = threadQuery.data.messages.find((msg: any) => msg.type === 'user');
+        if (firstUserMessage?.content) {
+          threadName = firstUserMessage.content.substring(0, 50);
+          if (firstUserMessage.content.length > 50) threadName += '...';
+        }
+      }
+
+      // Get project information
+      const projectId = threadQuery.data?.project_id || null;
+      const projectName = projectQuery.data?.name || '';
+
+      return {
+        threadId,
+        name: threadName,
+        projectId,
+        projectName,
+        agentRun: runningAgentRun,
+        isLoading,
+        error: hasError,
+      };
+    });
+  }, [runningThreadIds, threadQueries, agentRunQueries, projectQueries]);
+
+  const isLoadingThreads = threadQueries.some(q => q.isLoading) || agentRunQueries.some(q => q.isLoading) || projectQueries.some(q => q.isLoading);
+
+  const handleClose = () => {
+    onOpenChange(false);
+  };
+
+  const handleThreadStopped = () => {
+    // No need to close the dialog - the queries will automatically refetch
+    // and update the UI to show the current running state
+  };
+
+  return (
+    <Dialog open={open} onOpenChange={onOpenChange}>
+      <DialogContent className="sm:max-w-md">
+        <DialogHeader>
+          <div className="flex items-center gap-2">
+            <div className="flex h-8 w-8 items-center justify-center rounded-full bg-amber-100 dark:bg-amber-900">
+              <AlertTriangle className="h-4 w-4 text-amber-600 dark:text-amber-400" />
+            </div>
+            <DialogTitle className="text-lg font-semibold">
+              Parallel Runs Limit Reached
+            </DialogTitle>
+          </div>
+          <DialogDescription className="text-sm text-muted-foreground">
+            You've reached the maximum parallel agent runs allowed.
+          </DialogDescription>
+        </DialogHeader>
+
+        <div className="space-y-4">
+          {(runningThreadIds.length > 0 || runningCount > 0) && (
+            <div className="space-y-3">
+              <div className="flex items-center gap-2">
+                <h4 className="text-sm font-medium">Currently Running Agents</h4>
+              </div>
+              
+              {isLoadingThreads ? (
+                <div className="flex items-center justify-center py-4">
+                  <Loader2 className="h-4 w-4 animate-spin" />
+                  <span className="ml-2 text-sm text-muted-foreground">Loading threads...</span>
+                </div>
+              ) : runningThreadIds.length === 0 ? (
+                <div className="text-center py-4 text-sm text-muted-foreground">
+                  <p>Found {runningCount} running agents but unable to load thread details.</p>
+                  <p className="text-xs mt-1">Thread IDs: {JSON.stringify(runningThreadIds)}</p>
+                </div>
+              ) : (
+                <div className="space-y-2 max-h-64 overflow-y-auto">
+                  {runningThreadsInfo.slice(0, 5).map((threadInfo) => (
+                    <RunningThreadItem
+                      key={threadInfo.threadId}
+                      threadInfo={threadInfo}
+                      onThreadStopped={handleThreadStopped}
+                    />
+                  ))}
+                  
+                  {runningThreadsInfo.length > 5 && (
+                    <div className="text-center py-2">
+                      <Badge variant="outline" className="text-xs">
+                        +{runningThreadsInfo.length - 5} more running
+                      </Badge>
+                    </div>
+                  )}
+                </div>
+              )}
+            </div>
+          )}
+
+          <Separator />
+
+          <div className="space-y-3">
+            <h4 className="text-sm font-medium">What can you do?</h4>
+            <ul className="space-y-2 text-sm text-muted-foreground">
+              <li className="flex items-start gap-2">
+                <div className="h-1.5 w-1.5 rounded-full bg-muted-foreground mt-2 flex-shrink-0" />
+                <span>Click the <Square className="h-3 w-3 inline mx-1" /> button to stop running agents</span>
+              </li>
+              <li className="flex items-start gap-2">
+                <div className="h-1.5 w-1.5 rounded-full bg-muted-foreground mt-2 flex-shrink-0" />
+                <span>Wait for an agent to complete automatically</span>
+              </li>
+            </ul>
+          </div>
+
+          <div className="flex justify-end gap-2 pt-2">
+            <Button variant="outline" onClick={handleClose}>
+              Got it
+            </Button>
+          </div>
+        </div>
+      </DialogContent>
+    </Dialog>
+  );
+};

frontend/src/hooks/react-query/dashboard/use-initiate-agent.ts

@@ -1,6 +1,6 @@
 'use client';
 
-import { initiateAgent, InitiateAgentResponse } from "@/lib/api";
+import { initiateAgent, InitiateAgentResponse, BillingError, AgentRunLimitError } from "@/lib/api";
 import { createMutationHook } from "@/hooks/use-query";
 import { handleApiSuccess, handleApiError } from "@/lib/error-handler";
 import { dashboardKeys } from "./keys";
@@ -19,6 +19,10 @@ export const useInitiateAgentMutation = createMutationHook<
       handleApiSuccess("Agent initiated successfully", "Your AI assistant is ready to help");
     },
     onError: (error) => {
+      // Let BillingError and AgentRunLimitError bubble up to be handled by components
+      if (error instanceof BillingError || error instanceof AgentRunLimitError) {
+        throw error;
+      }
       if (error instanceof Error && error.message.toLowerCase().includes("payment required")) {
         return;
       }
@@ -38,6 +42,12 @@ export const useInitiateAgentWithInvalidation = () => {
     },
     onError: (error) => {
       console.log('Mutation error:', error);
+      
+      // Let AgentRunLimitError bubble up to be handled by components
+      if (error instanceof AgentRunLimitError) {
+        throw error;
+      }
+      
       if (error instanceof Error) {
         const errorMessage = error.message;
         if (errorMessage.toLowerCase().includes("payment required")) {

frontend/src/hooks/react-query/threads/use-agent-run.ts

@@ -1,6 +1,6 @@
 import { createMutationHook, createQueryHook } from "@/hooks/use-query";
 import { threadKeys } from "./keys";
-import { BillingError, getAgentRuns, startAgent, stopAgent } from "@/lib/api";
+import { BillingError, AgentRunLimitError, getAgentRuns, startAgent, stopAgent } from "@/lib/api";
 
 export const useAgentRunsQuery = (threadId: string) =>
   createQueryHook(
@@ -29,6 +29,7 @@ export const useStartAgentMutation = () =>
     }) => startAgent(threadId, options),
     {
       onError: (error) => {
+        // Only silently handle BillingError - let AgentRunLimitError bubble up to be handled by the page component
         if (!(error instanceof BillingError)) {
           throw error;
         }

frontend/src/lib/api.ts

@@ -29,6 +29,35 @@ export class BillingError extends Error {
   }
 }
 
+// Custom error for agent run limit exceeded
+export class AgentRunLimitError extends Error {
+  status: number;
+  detail: { 
+    message: string;
+    running_thread_ids: string[];
+    running_count: number;
+  };
+
+  constructor(
+    status: number,
+    detail: { 
+      message: string;
+      running_thread_ids: string[];
+      running_count: number;
+      [key: string]: any;
+    },
+    message?: string,
+  ) {
+    super(message || detail.message || `Agent Run Limit Exceeded: ${status}`);
+    this.name = 'AgentRunLimitError';
+    this.status = status;
+    this.detail = detail;
+
+    // Set the prototype explicitly.
+    Object.setPrototypeOf(this, AgentRunLimitError.prototype);
+  }
+}
+
 export class NoAccessTokenAvailableError extends Error {
   constructor(message?: string, options?: { cause?: Error }) {
     super(message || 'No access token available', options);
@@ -707,6 +736,28 @@ export const startAgent = async (
         }
       }
 
+      // Check for 429 Too Many Requests (Agent Run Limit)
+      if (response.status === 429) {
+          const errorData = await response.json();
+          console.error(`[API] Agent run limit error starting agent (429):`, errorData);
+          // Ensure detail exists and has required properties
+          const detail = errorData?.detail || { 
+            message: 'Too many agent runs running',
+            running_thread_ids: [],
+            running_count: 0,
+          };
+          if (typeof detail.message !== 'string') {
+            detail.message = 'Too many agent runs running';
+          }
+          if (!Array.isArray(detail.running_thread_ids)) {
+            detail.running_thread_ids = [];
+          }
+          if (typeof detail.running_count !== 'number') {
+            detail.running_count = 0;
+          }
+          throw new AgentRunLimitError(response.status, detail);
+      }
+
       // Handle other errors
       const errorText = await response
         .text()
@@ -723,8 +774,8 @@ export const startAgent = async (
     const result = await response.json();
     return result;
   } catch (error) {
-    // Rethrow BillingError instances directly
-    if (error instanceof BillingError) {
+    // Rethrow BillingError and AgentRunLimitError instances directly
+    if (error instanceof BillingError || error instanceof AgentRunLimitError) {
       throw error;
     }
 
@@ -1513,6 +1564,54 @@ export const initiateAgent = async (
     });
 
     if (!response.ok) {
+      // Check for 402 Payment Required first
+      if (response.status === 402) {
+        try {
+          const errorData = await response.json();
+          console.error(`[API] Billing error initiating agent (402):`, errorData);
+          // Ensure detail exists and has a message property
+          const detail = errorData?.detail || { message: 'Payment Required' };
+          if (typeof detail.message !== 'string') {
+            detail.message = 'Payment Required'; // Default message if missing
+          }
+          throw new BillingError(response.status, detail);
+        } catch (parseError) {
+          // Handle cases where parsing fails or the structure isn't as expected
+          console.error(
+            '[API] Could not parse 402 error response body:',
+            parseError,
+          );
+          throw new BillingError(
+            response.status,
+            { message: 'Payment Required' },
+            `Error initiating agent: ${response.statusText} (402)`,
+          );
+        }
+      }
+
+      // Check for 429 Too Many Requests (Agent Run Limit)
+      if (response.status === 429) {
+          const errorData = await response.json();
+          console.error(`[API] Agent run limit error initiating agent (429):`, errorData);
+          // Ensure detail exists and has required properties
+          const detail = errorData?.detail || { 
+            message: 'Too many agent runs running',
+            running_thread_ids: [],
+            running_count: 0,
+          };
+          if (typeof detail.message !== 'string') {
+            detail.message = 'Too many agent runs running';
+          }
+          if (!Array.isArray(detail.running_thread_ids)) {
+            detail.running_thread_ids = [];
+          }
+          if (typeof detail.running_count !== 'number') {
+            detail.running_count = 0;
+          }
+          throw new AgentRunLimitError(response.status, detail);
+      }
+
+      // Handle other errors
       const errorText = await response
         .text()
         .catch(() => 'No error details available');
@@ -1522,9 +1621,7 @@ export const initiateAgent = async (
         errorText,
       );
     
-      if (response.status === 402) {
-        throw new Error('Payment Required');
-      } else if (response.status === 401) {
+      if (response.status === 401) {
         throw new Error('Authentication error: Please sign in again');
       } else if (response.status >= 500) {
         throw new Error('Server error: Please try again later');
@@ -1538,6 +1635,11 @@ export const initiateAgent = async (
     const result = await response.json();
     return result;
   } catch (error) {
+    // Rethrow BillingError and AgentRunLimitError instances directly
+    if (error instanceof BillingError || error instanceof AgentRunLimitError) {
+      throw error;
+    }
+
     console.error('[API] Failed to initiate agent:', error);
 
     if (

frontend/src/lib/error-handler.ts

@@ -1,5 +1,5 @@
 import { toast } from 'sonner';
-import { BillingError } from './api';
+import { BillingError, AgentRunLimitError } from './api';
 
 export interface ApiError extends Error {
   status?: number;
@@ -50,6 +50,10 @@ const extractErrorMessage = (error: any): string => {
     return error.detail?.message || error.message || 'Billing issue detected';
   }
 
+  if (error instanceof AgentRunLimitError) {
+    return error.detail?.message || error.message || 'Agent run limit exceeded';
+  }
+
   if (error instanceof Error) {
     return error.message;
   }
@@ -85,6 +89,9 @@ const shouldShowError = (error: any, context?: ErrorContext): boolean => {
   if (error instanceof BillingError) {
     return false;
   }
+  if (error instanceof AgentRunLimitError) {
+    return false;
+  }
 
   if (error?.status === 404 && context?.resource) {
     return false;