🤖 Active Agents

Researcher Agent
Status: Synthesizing web search results

Coder Agent
Status: Generating Python service module

Executor Agent
Status: Standby

Crypto Agent
Status: Monitoring Base mempool

Monitor Agent
Status: Health checks running

📊 System Metrics

Tasks Processed (24h) 1,247

Queue Depth 23

Avg Response Time 1.2s

Success Rate 98.7%

Active Tools 12

📋 Task Queue

Task #3561: Research AGI safety grants
Agent: Researcher | Priority: 8

Task #3562: Generate MCP server code
Agent: Coder | Priority: 7

Task #3563: Monitor Base DEX for arbitrage
Agent: Crypto | Priority: 9

Task #3564: Execute paper simulation
Agent: Executor | Priority: 6

🔧 MCP Tool Registry

🔍 web_search

⚙️ execute_code

⛓️ blockchain_query

🌐 browser_automation

📄 file_operations

📊 data_analysis

🔐 crypto_operations

📝 document_generation

🗄️ database_query

📧 email_handler

🤖 ai_inference

📈 metrics_collector

System Architecture

User Intent

→

Marcus Controller

→

Task Router

→

Agent Pool

→

MCP Tools

→

Results

Key Features

Intent Decomposition: Groq Llama 3.3-70b breaks complex tasks into DAGs
Dynamic Routing: BullMQ + Redis Streams for async agent coordination
MCP Protocol: Standardized tool discovery and execution
State Management: PostgreSQL/SQLite tracking all task states
Auto-scaling: Agent pool expands based on queue depth
Health Monitoring: Real-time agent heartbeat tracking

Quick Start

# Clone repository
git clone https://code.broke2builtai.com/marcus-agent-orchestrator
cd marcus-agent-orchestrator

# Install dependencies
npm install

# Set environment variables
export GROQ_API_KEY=your_key_here
export REDIS_URL=redis://localhost:6379

# Start Marcus
npm run start

# Submit an intent via API
curl -X POST http://localhost:3000/api/intent \
  -H "Content-Type: application/json" \
  -d '{"intent": "Research AGI grants and generate application", "context": {}}'

⚡ MARCUS AGENT ORCHESTRATOR ⚡