Two-Agent Architecture

Security and Scalability Combined

Stavily’s Two-Agent Architecture solves the challenge of maintaining enterprise-grade security while achieving unlimited scalability. By separating monitoring (sensors) from execution (actions), we create a system where security and performance work together rather than against each other.

Key Security Benefits

Sensor Agent Security

Sensor agents can only monitor and report events as they cannot execute commands. If compromised, they provide no access to your systems, allowing safe investigation and response.

Action Agent Containment

Action agents perform only pre-approved tasks with minimal permissions. Any issues are contained, with automatic revocation preventing widespread damage.

Global Scale Security

Deploy sensor agents worldwide while maintaining centralized security controls. Security scales with your infrastructure, not against it.

How the Architecture Works

1. Monitor Silently: Sensor agents observe systems without making changes.
2. Report Securely: Encrypted communication sends data to the orchestrator.
3. Validate Requests: The orchestrator verifies every action request.
4. Execute Minimally: Action agents perform only approved, limited tasks.

Screenshot showing the agent management interface with sensor and action agent configurations.

Security Implementation

Defense in Depth

Traditional systems protect one large target, where any breach affects everything. Our two-agent approach creates multiple independent security layers, containing incidents to specific components.

Zero-Trust Principles

• Agent Verification: Every agent has cryptographic identity with continuous validation.
• Encrypted Communications: End-to-end encryption protects all data in transit.
• Complete Audit Trail: Every action is logged from trigger to completion.
• Minimal Permissions: Least privilege access with just-in-time authorization.

Security by Design

In today’s threat landscape, basic security measures are insufficient. Two-Agent Architecture provides the enterprise-grade protection required while maintaining operational flexibility.

Breach Impact Mitigation

• Sensor Compromise: No system access, monitoring only.
• Network Attacks: Encrypted channels with mutual authentication.
• Action Agent Issues: Pre-approved tasks with automatic revocation.
• Orchestrator Security: Isolated component with additional controls.

Scalability Features

Scale monitoring and execution separately based on your needs. Deploy thousands of lightweight sensor agents globally while strategically placing action agents where execution is required.

Global Architecture

graph LR
    subgraph "Worldwide Sensors"
        S1[Sensor New York] --> O[Orchestrator]
        S2[Sensor London] --> O
        S3[Sensor Tokyo] --> O
        S4[Sensor Sydney] --> O
    end

    subgraph "Strategic Actions"
        O --> A1[Action US East]
        O --> A2[Action EU West]
        O --> A3[Action Asia Pacific]
    end

    subgraph "Security Layers"
        SL1[Identity Verification] --> O
        SL2[Encryption Gateway] --> O
        SL3[Audit Logger] --> O
    end

Performance Optimization

• Efficient Sensors: Minimal CPU and memory usage (sub-1% CPU).
• Targeted Actions: Deploy execution capacity only where needed.
• Smart Networking: Optimal traffic routing with local caching.
• Cost Efficiency: Pay for actual usage and scale as required.

Deployment Options

• Cloud-Native: Kubernetes and containerized environments. Recommended
• On-Premises: Air-gapped and regulated environments.
• Hybrid: Seamless integration of cloud and on-premises systems.
• Edge Computing: Support for IoT, retail, and manufacturing deployments.