Plan before change
Expose intended resources, steps, locks, compatibility, and risk before assigning work.
Product
One controlled operating path for the Kubernetes lifecycle.
ClusterPilot standardizes provider infrastructure, cluster changes, add-ons, validation and evidence in a customer-hosted execution model.
Gate every boundary
Keep authorization, capability maturity, provider readiness, agent compatibility, and approval separate.
Preserve the truth
Project immutable history, stable errors, artifacts, checksums, audit, and explicit cleanup state.
Real product workflow
Connect the provider. Build the cluster. Keep the proof.
ClusterPilot keeps provider readiness, infrastructure state and Kubernetes operations connected across every handoff.
01 · Connect
Validate a provider account
Register write-only credentials, validate the provider boundary, and continue only when the account is ready for infrastructure.
02 · Provision
Plan role-based infrastructure
Review provider-safe defaults, node roles and the deterministic plan before infrastructure provisioning starts.
03 · Operate
Follow every operation
Track ordered lifecycle and add-on work with status, actor, duration, stable failure context and an explicit recovery path.
Operational depth without hidden execution.
Every capability follows explicit admission, immutable inputs, a compatible execution boundary, and evidence contracts.
Core contract
Deterministic OperationRuns
Critical actions become durable runs with explicit plans, ordered steps, assignments, events, bounded logs, artifacts, cancellation, retry, and recovery context.
Open documentation →Stable providers / gated
Provider infrastructure
Model accounts, validation, discovery, networks, machines, volumes, firewalls, quotas, and provider failures without mixing cloud calls into Kubernetes semantics.
Open documentation →Production-gated
Cluster lifecycle
Apply the selected OS hardening profile, prepare supported Linux nodes, bootstrap kubeadm, install Calico, scale, upgrade one minor, replace, inspect, and delete under explicit gates.
Open documentation →Production-gated
Add-ons and health
Resolve immutable artifact locks, validate dependencies, install curated or custom add-ons, and require semantic health evidence—not merely running pods.
Open documentation →Candidate / gated
Conformance evidence
Orchestrate pinned Sonobuoy quick, non-disruptive, or certified workflows with admission, privilege, artifacts, immutable results, and cleanup.
Open documentation →Production-gated
Release and security controls
Fail closed on unsafe production configuration, require immutable images and signed metadata, redact sensitive values, and preserve authorization and audit.
Open documentation →Built for the day after provisioning
See success, failure and governance in the same operating model.
A platform earns trust when operators can understand why work stopped, what is safe next and who controls the boundary.
How ClusterPilot fits
Use your tools. Standardize the operation between them.
ClusterPilot is a workflow and evidence layer, not a claim to replace every infrastructure or delivery system.
Terraform / IaCDeclares and provisions infrastructure
ClusterPilot adds lifecycle admission, agent execution state, recovery and evidence across infrastructure and Kubernetes.
CI/CD & GitOpsBuilds and delivers applications
ClusterPilot governs cluster platform operations and can install approved delivery add-ons without becoming the application pipeline.
Managed KubernetesProvides a provider-operated cluster service
ClusterPilot is customer-hosted and targets repeatable operations across customer-controlled provider and host boundaries.
Scripts & runbooksCapture expert procedures
ClusterPilot turns approved procedures into planned, stateful, observable and auditable OperationRuns.
Four-layer operating model
Clear ownership at every boundary.
The control plane decides and records. The agent executes bounded compatible work. External systems remain explicit dependencies rather than hidden platform internals.
1ExperienceWeb UI · API · automation
Intent, approval, visibility
2ControlAuth · plans · gates · workers
Policy and orchestration
3ExecutionCompatible lifecycle agents
Bounded provider and host work
4EvidencePostgreSQL · artifacts · audit
Durable state and proof
Designed to fail closed
Know why work stopped—and what is safe next.
Stable ProblemDetails codes, explicit next actions, operation state, dependency posture, and integrity-checked evidence replace vague “something failed” outcomes.
- No hidden mutation on read paths
- No tag-only release fallback
- No credential material in operation options
- No retry without replay and conflict controls
Talk to ClusterPilot
See whether ClusterPilot’s operating model fits your environment.
Bring us one high-friction lifecycle workflow. We will map it to the current product boundary and a useful technical evaluation.
hello@clusterpilot.de