The Deployment Stack handles the packaging, deployment and execution of FRAKTIΛ agents in real-world environments. It abstracts infrastructure into a no-code, container-based deployment pipeline, ensuring high availability, observability and rollback support for any agent.

Each deployment is powered by lightweight containers that include the runtime, memory interface and agent-specific logic.

Architecture Overview

Agent Deployment Spec (JSON Example)

"deployment": {
  "replicas": 1,
  "region": "eu-central",
  "expose": {
    "rest": true,
    "websocket": true,
    "voice": true
  },
  "resources": {
    "cpu": "300m",
    "memory": "512Mi"
  },
  "autoRestart": true,
  "watchdog": {
    "enabled": true,
    "threshold": 3,
    "action": "reboot"
  }
}

Supported Features

✦ One-click deploy from Zero-Code UI.
✦ Cloud + Edge support. (Kubernetes-native or serverless mode)
✦ Service discovery for inter-agent communication.
✦ Versioned rollback and snapshot recovery.
✦ Audit trail for all deploy/update events.

Monitoring and Observability

Deployed agents are trackable through the Agent Dashboard, including:

✦ Memory usage and CPU consumption.
✦ API call frequency. (Add-On usage)
✦ Response time latency.
✦ Model token usage.
✦ Trigger & fallback paths.
✦ Uptime and error logs.

Security Layer

✦ TLS encryption for all endpoints.
✦ Agent tokens scoped to environment.
✦ Access logs + permission enforcement via dashboard.
✦ Optional private-mode deployment for closed environments.

Why It Matters

Without needing any DevOps knowledge, users can:

✦ Launch agents on scalable, isolated infrastructure.
✦ Maintain uptime across upgrades.
✦ Debug live behavior.
✦ Delegate execution across teams and roles.