Your infrastructure is
    already a database.
    It's time to query it.

    Orbi is a SQL-like DSL for managing infrastructure across 10 providers — Azure, AWS, Cloudflare, GitHub, Kubernetes, Docker, VMware, and more — with live queries against real cluster state.

    $ curl -fsSL https://raw.githubusercontent.com/epyphite/orbi/main/install.sh | sh
    Get Started GitHub
    A real platform: 108,511 lines of Terraform/Helm → 1,235 lines of Orbi. See the comparison →

    The query you can't run anywhere else.

    orbi shell
    $ orbi "SELECT name, namespace, status, restarts FROM k8s_pods WHERE status = 'CrashLoopBackOff';"

┌─────────────────────┬────────────┬──────────────────┬──────────┐
│ name                │ namespace  │ status           │ restarts │
├─────────────────────┼────────────┼──────────────────┼──────────┤
│ api-worker-7f9b6     │ prod       │ CrashLoopBackOff │       14 │
│ billing-sync-2d1c    │ prod       │ CrashLoopBackOff │        8 │
│ notif-dispatcher-xy  │ staging    │ CrashLoopBackOff │        3 │
└─────────────────────┴────────────┴──────────────────┴──────────┘
3 rows

$ orbi "SELECT name, replicas, ready_replicas FROM k8s_deployments WHERE ready_replicas < replicas;"

┌──────────────────┬──────────┬─────────────────┐
│ name             │ replicas │ ready_replicas  │
├──────────────────┼──────────┼─────────────────┤
│ api-worker       │        5 │               2 │
│ billing-sync     │        3 │               0 │
└──────────────────┴──────────┴─────────────────┘
2 rows

$ orbi "CREATE RESOURCE 'cf_dns_record' id='api' zone='example.com' type='A' content='1.2.3.4';"
✓ Created cf_dns_record 'api' in zone example.com → 1.2.3.4

$

    This is orbi querying a live Kubernetes cluster.
    Not a static snapshot. Not a Terraform state file. The actual pods, right now.

    See it run.

    Real terminal recordings — no scripts, no mocks, no edits.

    Resource Discovery

    Orbi discovers and queries live infrastructure across providers — every resource as a row.

    Cost Analysis

    Query infrastructure costs across providers — spot waste, compare environments, track spend.

    -- Which pods are crashing across all namespaces?
SELECT name, namespace, restarts
FROM k8s_pods
WHERE status = 'CrashLoopBackOff';

-- Which deployments haven't caught up to their desired replica count?
SELECT name, replicas, ready_replicas
FROM k8s_deployments
WHERE ready_replicas < replicas;

-- Which firewall rules block traffic from Tor exit nodes?
SELECT id, expression, action
FROM resources
WHERE resource_type = 'cf_firewall_rule'
  AND expression LIKE '%tor%';

    SQL against your cloud.

    kubectl get pods | grep CrashLoopBackOff works, until you need to sort by restart count, filter by namespace, and join against deployment names.

    Orbi treats every resource — Kubernetes pods, Azure databases, AWS VPCs, Cloudflare DNS records — as rows in a table.

    Query them with WHERE clauses. Filter by tags. Sort by status.

    No other tool does this.

    # why

    We built Orbi after the 400th time we wrote a bash script to find drift between our Terraform state and reality. If infrastructure is data, we thought, why can't we query it? Turns out you can — and once you can, everything else about managing infrastructure gets easier.

    Declarative, not imperative

    Describe what you want. Orbi figures out how. IF NOT EXISTS makes every statement idempotent. Re-running a file is safe.

    CREATE IF NOT EXISTS RESOURCE 'postgres'
  id = 'prod-db'
  version = '16'
  storage_gb = 64;

    Queryable, not static

    Your registry is SQLite. Your cloud state is a view. SELECT anything. Join anything. Export anything.

    SELECT resource_type, COUNT(*)
FROM resources
WHERE status = 'pending'
GROUP BY resource_type;

    One language. Ten providers.

    Azure, AWS, Cloudflare, GitHub, Kubernetes, Docker, VMware, GCP, Firecracker, and SSH. Same CREATE/SELECT/DESTROY. Different ON PROVIDER.

    CREATE RESOURCE 'postgres' id = 'db'
  ON PROVIDER 'azure-prod';

CREATE RESOURCE 'rds_postgres' id = 'dr'
  ON PROVIDER 'aws-dr';

    59 resource types across 10 providers.

    Azure
    16 resources

    PostgreSQL, AKS, networking, container apps, DNS, ACR

    AWS
    23 resources

    RDS, EKS, VPC, S3, MSK, ElastiCache, IAM, SES, Backup

    Cloudflare
    4 resources

    DNS, firewall, page rules, zones

    GitHub
    6 resources

    Repos, rulesets, secrets, workflow files

    Kubernetes
    6 + live query

    Deployments, services, LIVE pod queries

    Docker
    4 resources

    Containers, networks, volumes, Compose stacks

    VMware
    2+ resources

    VMs, snapshots — vSphere and Workstation/Fusion

    Firecracker
    MicroVMs

    Local dev + prod edge compute

    Plus 9 credential backends: env, file, HashiCorp Vault, AWS Secrets Manager, GCP Secret Manager, Azure Key Vault, 1Password, SOPS, Kubernetes Secrets.

    -- Kubernetes deployment from zero
ADD IF NOT EXISTS PROVIDER id = 'prod'
  type = 'kubernetes' auth = 'env:KUBECONTEXT';

CREATE RESOURCE 'k8s_namespace' id = 'api';

CREATE RESOURCE 'k8s_deployment' id = 'api'
  namespace = 'api'
  image = 'myregistry.azurecr.io/api:v1.2.3'
  replicas = 3
  port = 8080
  ON PROVIDER 'prod';

CREATE RESOURCE 'k8s_service' id = 'api'
  namespace = 'api'
  type = 'LoadBalancer'
  port = 80 target_port = 8080
  ON PROVIDER 'prod';

    Try it in 30 seconds.

    No credentials required. Simulate mode runs the full DSL against realistic fake clouds. Point it at real infrastructure when you're ready.

    # Install
curl -fsSL https://raw.githubusercontent.com/epyphite/orbi/main/install.sh | sh

# Try it — no credentials needed
orbi --simulate "CREATE RESOURCE 'postgres' id = 'test-db' version = '16';"

# Interactive shell
orbi shell

    No spam. Unsubscribe anytime. We'll email you when new providers ship.