feat: AI Swarm Mode - Multi-agent parallel task execution

[Suhaib3100]

Summary

AI Swarm Mode enables a master agent to spawn and orchestrate multiple worker agents in separate browser windows, parallelizing complex multi-step tasks.

Use Case

User: "Research and compare the top 5 CRM solutions"

Master Agent (Coordinator)
├── Decomposes task into 5 parallel subtasks
├── Spawns 5 worker windows
├── Monitors progress
└── Synthesizes final report
    │
    ├── Worker 1: Research Salesforce
    ├── Worker 2: Research HubSpot
    ├── Worker 3: Research Pipedrive
    ├── Worker 4: Research Zoho
    └── Worker 5: Research Monday

Key Components

Phase 1: Foundation

• Swarm types and constants
• SwarmRegistry (active swarm tracking)
• WorkerLifecycleManager (spawn/terminate)
• SwarmMessagingBus (inter-agent communication)

Phase 2: Communication

• Heartbeat protocol
• Worker health monitoring
• Progress tracking
• Retry logic with exponential backoff

Phase 3: LLM Decomposition

• TaskPlanner (auto-decompose tasks)
• Worker count estimation
• Output schema generation

Phase 4: Aggregation

• ResultAggregator
• Conflict resolution
• LLM-based synthesis

Phase 5: API & Integration

• /swarm API routes
• SSE streaming for status
• Agent SDK extensions

Benefits

• 5x+ speedup for parallelizable tasks
• Robust failure recovery
• Real-time progress monitoring
• Resource isolation per worker