API Overview
Complete OpenAPI documentation for k0rdent AI Atlas, Arc, and shared services APIs
Overview
Draft: This documentation is currently a work in progress and subject to change.
This document provides an overview of the k0rdent API shape, covering Atlas (Provider Console), Arc (Customer AI Console), and shared services like Auth.
API Docs Variants
We're experimenting with different ways to present the API. Choose the format that works best for you.
📍 This Docs Site
API + Product Docs combined for the k0rdent AI platform
⚖️ Scalar
🔑 Vanilla Scalar documentation UI
📚 Redocly
🔑 Vanilla Redocly documentation UI
📖 Swagger UI
🔑 Vanilla Swagger UI documentation UI
Welcome to the k0rdent AI documentation
k0rdent AI is a Kubernetes-native platform for managing AI infrastructure at scale. Built for Neoclouds and enterprises, it provides a composable, template-driven approach to deploying and operating Kubernetes clusters and AI services across distributed environments.
What k0rdent AI solves
Modern AI infrastructure faces three core challenges:
Tech stack complexity: Divergent tooling, APIs, and security controls across providers and regions make it hard to standardize operations as hardware and platforms proliferate.
Operational efficiency: Without centralized visibility, cluster health, state, cost, and risk drift across environments, creating instability and compliance gaps.
Customer experience: Slow time-to-value onboarding new GPU capacity and services, difficulty maximizing utilization while meeting strict tenant requirements, and hard multi-tenancy raise the bar for reliability and isolation.
k0rdent AI addresses these challenges through Kubernetes-native composability and multi-cluster automation. Infrastructure and services are defined as templates and reconciled continuously into running environments, enabling repeatable rollout across many clusters and sites with controlled drift and upgrades.
Key capabilities
Composable infrastructure: Define infrastructure and services as templates, reconcile them across clusters and sites GPU efficiency: GPU partitioning, topology-aware scheduling, and allocation strategies to improve utilization Hardware-enforced multi-tenancy: Tenant isolation across GPU allocation, virtualization, and network boundaries Integrated observability and FinOps: Centralized monitoring and cost reporting for operational control and tenant-level consumption reporting Workload convergence: Support for VM and container workloads under one operational model
Platform components
k0rdent AI consists of three main components:
k0rdent AI Cluster Manager (KCM): Deployment and lifecycle management of Kubernetes clusters, including configuration, updates, and other CRUD operations.
k0rdent AI State Manager (KSM): Installation and lifecycle management of deployed services. KSM leverages Project Sveltos for an increasing amount of functionality.
k0rdent AI Observability and FinOps (KOF): Cluster and beach-head services monitoring, events and log management, with integrated cost reporting and tenant-level consumption tracking.
API Endpoint Reference
All endpoints are served from https://api.k0rdent.ai. Visibility is controlled per-endpoint via the x-visibility extension: internal (provider/Atlas operations only) or public (customer/Arc-facing).
Compute
Region-scoped compute resources under /v1/regions/{region}/projects/{project}/compute/.
| Endpoint | Purpose |
|---|---|
/v1/regions/{region}/projects/{project}/compute/clusters | Kubernetes cluster deployments |
/v1/regions/{region}/projects/{project}/compute/clusters/{clusterId}/kubeconfigs | Kubeconfig issuance and management |
/v1/regions/{region}/projects/{project}/compute/instances | VM and bare metal instance lifecycle |
/v1/regions/{region}/projects/{project}/compute/networks | Network configuration and topology |
/v1/regions/{region}/projects/{project}/compute/subnets | Subnet management |
/v1/regions/{region}/projects/{project}/compute/addresses | IP address allocation and release |
Infrastructure
Provider-only (internal visibility) infrastructure resources.
| Endpoint | Purpose |
|---|---|
/v1/regions/{region}/projects/{project}/infrastructure/servers | Bare metal server lifecycle — enrollment, provisioning, state transitions |
Organizations & Projects
Global resources for tenant and project management.
| Endpoint | Purpose |
|---|---|
/v1/regions/global/organizations | Organization (tenant) management |
/v1/regions/global/organizations/{id}/invitations | Organization invitation management |
/v1/regions/global/projects | Project resource grouping |
Authentication
Token lifecycle and permission evaluation under /v1/regions/global/auth/.
| Endpoint | Purpose |
|---|---|
/v1/regions/global/auth/token | Mint access tokens (authorization_code, api_key, client_credentials) |
/v1/regions/global/auth/introspect | Token introspection (RFC 7662) — validate and decode token claims |
/v1/regions/global/auth/check | Evaluate permissions for a principal against actions and resources |
/v1/regions/global/auth/revoke | Token revocation (RFC 7009) |
IAM
Identity and access management resources under /v1/regions/global/iam/.
| Endpoint | Purpose |
|---|---|
/v1/regions/global/iam/users | User management and profile |
/v1/regions/global/iam/groups | User groups for team-based access control |
/v1/regions/global/iam/roles | RBAC role definitions |
/v1/regions/global/iam/policies | IAM policy bindings |
/v1/regions/global/iam/apikeys | API keys for programmatic access |
/v1/regions/global/iam/service-accounts | Machine-to-machine service accounts and credentials |
/v1/regions/global/iam/providers | Identity provider (OAuth/OIDC) configuration |
Future / TBD
The following endpoints are under discussion and have not been implemented yet.
| Endpoint | Purpose |
|---|---|
/v1/regions/global/audit | Audit logs |
/v1/regions/global/billing | Billing and consumption |
/v1/regions/global/analytics | Analytics |
/v1/regions/global/webhooks | Webhook subscriptions |
/v1/regions/{region}/projects/{project}/compute/instances (VMs) | Virtual machine lifecycle (in progress) |
/v1/regions/{region}/inference | Inference endpoint lifecycle |
/v1/regions/{region}/training | Training job lifecycle |
API Changelog
For a full history of API endpoint changes and iterations, see the Changelog.
Authentication
Versioning
Common Structures
Errors
Standards
Clusters
Kubeconfigs
Instances
Networks
Addresses
Servers
Organizations
Projects
Token
Introspect
Check
Revoke
Users
Roles
Groups
Policies
Apikeys
Serviceaccounts
Providers
Changelog
Last updated on
How is this guide?