Zillion Agent

{
  "protocolVersion": "0.3.0",
  "name": "Zillion Agent",
  "description": "An autonomous AI infrastructure specialist that manages compute end-to-end \u2014 from bare-metal and VMs through Kubernetes, AI model serving, and day-2 OAM. Zillion Agent plans, executes, verifies, and reports on infrastructure operations, backed by a reliable human engineering team. We help you manage your AI infrastructure end-to-end. For onboarding, demos, or partnerships, contact joel@zillion.network.",
  "version": "1.0.0",
  "url": "wss://app-backend.zillion.network",
  "preferredTransport": "JSONRPC",
  "provider": {
    "organization": "Zillion Network",
    "url": "wss://app-backend.zillion.network"
  },
  "capabilities": {
    "streaming": true,
    "pushNotifications": true,
    "stateTransitionHistory": true
  },
  "defaultInputModes": [
    "text/plain",
    "application/json"
  ],
  "defaultOutputModes": [
    "text/plain",
    "text/markdown",
    "application/json"
  ],
  "securitySchemes": {
    "zillionAccount": {
      "type": "http",
      "scheme": "bearer",
      "description": "Authenticate with a Personal Access Token (PAT) sent as an HTTP bearer token. To obtain one: (1) register an account on the portal at https://app.zillion.network and verify your email \u2014 registration is portal-only, there is no terminal/API sign-up; (2) sign in, open the Tokens / API Tokens page, and create a token. The PAT is shown once, so copy it immediately; it is a permanent credential tied to your account. For access questions or partnerships, contact joel@zillion.network."
    }
  },
  "security": [
    {
      "zillionAccount": []
    }
  ],
  "skills": [
    {
      "id": "kubernetes-cluster-lifecycle",
      "name": "Kubernetes Cluster Lifecycle",
      "description": "Provision complete Kubernetes clusters (control-plane + workers), add or remove nodes, and tear clusters down cleanly via a hardened, repeatable engine with post-deployment verification.",
      "tags": [
        "kubernetes",
        "k8s",
        "cluster",
        "provisioning",
        "infrastructure",
        "orchestration"
      ],
      "examples": [
        "Stand up a 3-node Kubernetes cluster with one control node and two workers.",
        "Add a GPU worker node to my existing cluster.",
        "Tear down the staging cluster cleanly."
      ]
    },
    {
      "id": "gpu-enablement",
      "name": "GPU Enablement",
      "description": "Enable end-to-end GPU support on new or existing clusters and servers: NVIDIA drivers, CUDA, and the container device runtime.",
      "tags": [
        "gpu",
        "nvidia",
        "cuda",
        "accelerator",
        "ai-infrastructure"
      ],
      "examples": [
        "Enable GPU support on my worker nodes.",
        "Install CUDA and the NVIDIA device plugin on this cluster."
      ]
    },
    {
      "id": "ai-model-serving",
      "name": "AI Model Serving (vLLM)",
      "description": "Deploy and serve large language models with vLLM onto a cluster or a single server's container runtime, validating model availability before deployment and serving health afterward.",
      "tags": [
        "vllm",
        "llm",
        "model-serving",
        "inference",
        "ai"
      ],
      "examples": [
        "Deploy an available LLM with vLLM on my cluster.",
        "Serve a model on a single GPU server using Docker."
      ]
    },
    {
      "id": "server-operations",
      "name": "Secure Server Operations",
      "description": "Operate authorized servers over key-based SSH with a read-first, OS-aware workflow: gather context, propose a solution, get approval, then execute. Network-breaking changes are blocked by design.",
      "tags": [
        "ssh",
        "linux",
        "sysadmin",
        "operations",
        "remote-execution"
      ],
      "examples": [
        "Check disk usage across my servers and clean up the largest logs.",
        "Investigate why this service keeps restarting and propose a fix."
      ]
    },
    {
      "id": "proactive-monitoring-self-healing",
      "name": "Proactive Monitoring & Self-Healing",
      "description": "Always-on, agentless health monitoring that doesn't just alert \u2014 it acts. When CPU, RAM, disk, GPU utilization, temperatures, or reachability cross your thresholds, Zillion Agent automatically investigates the root cause, attempts safe self-healing remediation, and pushes a clear, AI-written analysis to Telegram and Slack so your team wakes up to answers, not just red dashboards.",
      "tags": [
        "proactive-monitoring",
        "self-healing",
        "auto-remediation",
        "alerting",
        "observability",
        "incident-response",
        "telegram",
        "slack",
        "aiops"
      ],
      "examples": [
        "Watch all my servers and automatically diagnose and try to fix anything that breaches my thresholds.",
        "If a server goes offline, investigate why and tell me what you found and what you did.",
        "Alert my team on Slack with root-cause analysis whenever GPU temperature spikes."
      ]
    },
    {
      "id": "network-diagnostics",
      "name": "Network Diagnostics (OAM)",
      "description": "Globally distributed ping, traceroute, and DNS testing from multiple regions simultaneously to diagnose latency, packet loss, routing problems, and geo-DNS behavior.",
      "tags": [
        "network",
        "ping",
        "traceroute",
        "dns",
        "oam",
        "diagnostics"
      ],
      "examples": [
        "Ping my endpoint from China, the US, and Japan and summarize latency and packet loss.",
        "Run a traceroute from multiple regions to find where the route degrades."
      ]
    },
    {
      "id": "agent-managed-sandboxes",
      "name": "Agent-Managed Sandboxes at Scale",
      "description": "Zillion Agent spins up and tears down kernel-isolated microVM sandboxes on demand \u2014 fully managed for you. Run untrusted code, builds, and experiments in disposable Linux environments that never touch the host, scale out to massive fleets of sandboxes in parallel, and pick the right flavor for the job from lightweight to memory- and compute-heavy. The agent owns the full lifecycle: spawn, run, scale, and destroy.",
      "tags": [
        "sandbox",
        "microvm",
        "on-demand",
        "agent-managed",
        "scalable",
        "flavors",
        "isolation",
        "code-execution",
        "security"
      ],
      "examples": [
        "Spin up a sandbox, run this build, and tear it down when it's done.",
        "Launch 50 isolated sandboxes in parallel to run my test matrix.",
        "Use a high-memory sandbox flavor for this large workload."
      ]
    },
    {
      "id": "fleet-provisioning",
      "name": "Fleet & Bare-Metal Provisioning",
      "description": "B2B partner-grade fleet automation: provision, configure, and manage servers programmatically with a jobs queue, artifacts, and full audit logs.",
      "tags": [
        "bare-metal",
        "fleet",
        "provisioning",
        "automation",
        "jobs",
        "audit"
      ],
      "examples": [
        "Provision a new bare-metal server and put it into the available pool.",
        "Run a provisioning job and wait for it to complete."
      ]
    },
    {
      "id": "scheduled-automation",
      "name": "Scheduled Automation",
      "description": "Put recurring operations on auto-pilot. Schedule one-time or recurring tasks that wake Zillion Agent to run reports, health checks, and maintenance, with results delivered back to your channel automatically \u2014 no babysitting required.",
      "tags": [
        "scheduling",
        "automation",
        "auto-pilot",
        "reports",
        "maintenance"
      ],
      "examples": [
        "Every morning, check all servers and report anything above 80% disk.",
        "Send me a monthly cluster deployment summary on the first of each month."
      ]
    },
    {
      "id": "infrastructure-dashboards",
      "name": "Infrastructure Health Dashboards",
      "description": "Generate rich, interactive health-check and diagnostic dashboards so complex infrastructure state is instantly readable.",
      "tags": [
        "dashboard",
        "visualization",
        "health-check",
        "reporting"
      ],
      "examples": [
        "Show me a health dashboard for this cluster.",
        "Visualize the current resource usage across my fleet."
      ]
    }
  ]
}