Menu
Guided exploration of the HostedCluster create flow from API types through GKE provisioning
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Process GCP HCP meeting notes/transcripts into reviewed Jira updates — extracts action items, presents changes for approval, executes via MCP, and generates a Slack summary.
Use when working on GCP HCP-related code or design in any of the GCP HCP repositories: hypershift (platform/gcp/, GCP provider, GCP-related tests), gcp-hcp-infra (Terraform, ArgoCD, infrastructure), gcp-hcp-cli, cls-backend, cls-controller, or when creating or reviewing design decisions, or needing architectural context for GCP Hosted Control Planes on GKE. Accepts optional topic filter: networking, identity, observability, infrastructure, automation, governance, storage, operators, testing, incidents, slo.
Post a biweekly Agentic SDLC pilot update comment to GCP-579, gathering Jira and GitHub activity for the lookback period.
Use when adding a new GCP service account for Workload Identity Federation (WIF) in HyperShift and related GCP HCP repositories. Covers e2e plumbing, CI pipeline, API types, IAM bindings, CRD regeneration, CLS backend/controller propagation, and CLI updates across hypershift, openshift/release, gcp-hcp-infra, cls-backend, cls-controller, and gcp-hcp-cli.
Use when an implementation plan in gcp-hcp may have drifted from actual code, or when code may have drifted from the plan's design intent. Typically after a feature has landed or a PR has merged. Checks API surface, behavior, interfaces, and structure across repositories.
| name | walkthrough-hostedcluster-create |
| description | Guided exploration of the HostedCluster create flow from API types through GKE provisioning |
| user-invocable | false |
| disable-model-invocation | true |
Trace the end-to-end flow of creating a HostedCluster on GCP — from API definition through controller reconciliation to GKE cluster provisioning and status updates.
Start here. Find the HostedCluster API type definition in the hypershift repository. Show me:
HostedClusterSpec and HostedClusterStatus structuresHelp me understand: the API types are the contract. Everything downstream is driven by what's declared here.
Trace how HostedCluster creation requests are validated before they reach the controller. Find:
Help me understand: validation is the first line of defense. It catches misconfigurations before they become expensive reconciliation failures.
This is the core of the system. Find the HostedCluster controller and trace its reconcile function. Show me:
Help me understand: the reconcile loop is the engine. It continuously drives the actual state toward the desired state declared in the API types.
Trace from the controller into the GCP-specific provisioning code. Show me:
Help me understand: this layer bridges the Kubernetes abstraction to the cloud provider reality. The patterns here (credential management, API error handling, resource lifecycle) repeat across all GCP interactions.
Follow how provisioning status flows back to the HostedCluster object. Show me:
Help me understand: status is the feedback loop. It's how the system communicates progress, failures, and health — both to operators and to other controllers that depend on the HostedCluster being ready.
As you trace the flow, also point out these patterns when you encounter them:
Throughout the exploration, reference the design decisions documented in this repository (design-decisions/) to explain why certain architectural choices were made.