pr-workflow

name	PR Workflow
description	Clone kubernetes-manifests, branch, fix, validate, and open a PR. Use when: You've identified an issue in the kubernetes-manifests repo that needs fixing — resource limits, HelmRelease values, Flux kustomization config, namespace labels, or any manifest change. This is the standard workflow for making changes via pull request. Don't use when: The change is in openclaw-workspace repo (that's a direct push to main, not a PR to kubernetes-manifests). Don't use for diagnosing issues (use cluster-health, flux-ops, or storage-ops first, then come here to fix). Don't use when you're unsure what to change — diagnose first. Outputs: A pull request on rajsinghtech/kubernetes-manifests with validated changes, conventional commit message, and descriptive PR body.
requires	[]

PR Workflow

Routing

Use This Skill When

• You've diagnosed an issue and know the fix involves kubernetes-manifests
• Increasing resource limits for a crashing pod
• Fixing HelmRelease values or version bumps
• Fixing Flux kustomization config (paths, dependencies, substitution vars)
• Adding/modifying namespace labels or annotations
• Someone says "open a PR to fix this"

Don't Use This Skill When

• The change is to openclaw-workspace repo → commit directly to main
• You haven't diagnosed the issue yet → use cluster-health, flux-ops, or storage-ops first
• You're only inspecting/reading manifests → just use kubectl or clone and read
• The issue is with CI workflows → those live in openclaw-workspace, not kubernetes-manifests

Pre-Flight

Before starting, check for duplicate PRs:

gh pr list --repo rajsinghtech/kubernetes-manifests --state open

If an open PR already addresses this issue, comment on it instead of creating a new one.

Setup

# Always clean up stale clones first — leftover state from previous sessions causes confusion
rm -rf /tmp/k8s-manifests
git clone https://github.com/rajsinghtech/kubernetes-manifests.git /tmp/k8s-manifests
cd /tmp/k8s-manifests

⚠️ Always rm -rf before cloning. Stale clones from previous sessions will have the wrong branch, uncommitted changes, or outdated refs.

Branch Naming

Use conventional branch names:

• fix/ottawa-<description> — bug fix for Ottawa cluster
• fix/robbinsdale-<description> — bug fix for Robbinsdale
• fix/stpetersburg-<description> — bug fix for StPetersburg
• fix/infra-<description> — shared infrastructure fix
• feat/<description> — new feature or addition
• chore/<description> — maintenance, cleanup

git checkout -b fix/ottawa-coredns-memory-limit

Common Fix Templates

Resource Limits

# Increase memory limit for a crashing pod
resources:
  limits:
    memory: "512Mi"  # was 256Mi, OOMKilled
  requests:
    memory: "256Mi"

Flux Kustomization Config

# Fix dependency or path issues
spec:
  dependsOn:
    - name: infrastructure
  path: ./apps/media
  sourceRef:
    kind: GitRepository
    name: flux-system

HelmRelease Version Bump

spec:
  chart:
    spec:
      version: ">=1.2.3"  # bump from 1.2.2

HelmRelease Values Fix

spec:
  values:
    persistence:
      enabled: true
      storageClass: "ceph-block"  # was wrong class

Namespace Label/Annotation

metadata:
  labels:
    istio-injection: enabled  # add missing mesh label

Validate Before Committing

# Validate YAML syntax
yq . <changed-file.yaml> > /dev/null

# Kustomize build (if applicable)
kustomize build <path-to-kustomization-dir>/ > /dev/null

# Check for common mistakes
grep -rn '<<<' /tmp/k8s-manifests/  # merge conflicts
grep -rn 'TODO\|FIXME\|HACK' <changed-files>  # leftover markers

Commit and Push

git add <specific-files>
git commit -m "fix(ottawa): increase coredns memory limit to prevent OOM"
git push origin fix/ottawa-coredns-memory-limit

Commit message conventions:

• fix(<scope>): bug fix
• feat(<scope>): new feature
• chore(<scope>): maintenance
• Scope: cluster name (ottawa, robbinsdale, stpetersburg) or infra for shared

Create PR

gh pr create --repo rajsinghtech/kubernetes-manifests \
  --title "fix(ottawa): increase coredns memory limit to prevent OOM" \
  --body "$(cat <<'EOF'
## Problem
CoreDNS pods on Ottawa are OOMKilled with current 256Mi limit.

## Fix
Increased memory limit to 512Mi, request to 256Mi.

## Affected
- Cluster: talos-ottawa
- Namespace: kube-system
- Resource: deployment/coredns
EOF
)"

PR Body Template

## Problem
<What's broken and evidence (pod status, error message, alert)>

## Fix
<What was changed and why this specific value/config>

## Affected
- Cluster: <cluster name>
- Namespace: <namespace>
- Resource: <kind/name>

## Validation
- [ ] YAML syntax valid (`yq .`)
- [ ] Kustomize build passes (if applicable)
- [ ] No unrelated changes included

Report Back

After creating the PR, report:

[ottawa] PR opened: fix(ottawa): increase coredns memory limit
URL: https://github.com/rajsinghtech/kubernetes-manifests/pull/123
Files changed: apps/kube-system/coredns/deployment.yaml

Rules

• Never push to main — always branch and PR
• Never skip validation — kustomize build must pass
• One concern per PR — don't mix unrelated fixes
• Include context — PR body must explain the problem, fix, and affected resources
• Check for duplicates first — gh pr list --repo rajsinghtech/kubernetes-manifests --state open

Edge Cases

• Multiple clusters need the same fix: One PR per cluster, or one PR if the file is shared
• Fix requires SOPS changes: Dyson cannot modify SOPS — escalate to user
• Unsure about the right value: State your reasoning in the PR body and request review