apply.sh: kubeconfig merge + auto-apply mode; simplify README

[rowan-stein]

User request

• Add kubeconfig update to apply.sh using kubectl config view --merge --flatten; ask before updating in interactive mode.
• Add auto-apply mode to apply.sh: do not ask for domain/port; run Terraform with auto-apply flag(s); kubeconfig should be updated automatically.
• Rewrite root README.md: remove current content; keep only 3 sections:
  • Requirements: terraform, kubectl (optional)
  • Setup: Fast (chmod + run apply.sh), Manual (terraform init + terraform apply for each stack + kubeconfig update)
  • Usage: list exposed services with prebuilt links for default domain+port

Specification

apply.sh

1. Kubeconfig merge (post-apply)

• After all stacks apply successfully, optionally merge the generated kubeconfig at:

$(pwd)/stacks/k8s/.kube/agyn-local-kubeconfig.yaml

into:

$HOME/.kube/config

using:

KUBECONFIG="$KUBECONFIG:$HOME/.kube/config:$(pwd)/stacks/k8s/.kube/agyn-local-kubeconfig.yaml" \
  kubectl config view --merge --flatten

• Interactive mode (no auto-apply flag): prompt before modifying ~/.kube/config:
  • Merge k3d kubeconfig into ~/.kube/config? [Y/n]: (default Yes)
• Auto-apply mode: update kubeconfig automatically (no prompt).
• Implement the merge safely (don’t redirect output directly into an input file). Use a temp file + atomic mv (e.g. mktemp in ~/.kube/) or equivalent safe buffering.
• If kubectl is missing or the generated kubeconfig file is missing, print a warning and skip (non-fatal).

2. Auto-apply mode behavior

• Add/ensure a CLI flag (existing -y is OK) that:
  • runs Terraform with -input=false -auto-approve
  • does not prompt for domain/port
  • uses defaults when DOMAIN / PORT are unset
  • updates kubeconfig automatically (see above)

3. Domain/port resolution

• In interactive mode: if DOMAIN/PORT env vars are unset, prompt with defaults (existing behavior).
• In auto-apply mode: if DOMAIN/PORT env vars are unset, use defaults without prompting.

README.md (rewrite)

Replace the entire file with exactly 3 top-level sections:

1. Requirements

• List only:
  • terraform
  • kubectl (optional; only needed for kubeconfig merge / interacting with cluster)

2. Setup

• Fast: chmod +x apply.sh then run ./apply.sh (mention -y for auto-apply mode).
• Manual: show terraform init + terraform apply for each stack in order (stacks/k8s, stacks/system, stacks/routing, stacks/platform) + kubeconfig update step.

3. Usage

• List exposed services with prebuilt links for the defaults:
  • domain: agyn.dev
  • port: 2496
• Services to include (as defined by routing manifests):
  • Platform UI: https://agyn.dev:2496/
  • Platform API: https://agyn.dev:2496/api
  • Argo CD: https://argocd.agyn.dev:2496/
  • LiteLLM: https://litellm.agyn.dev:2496/
  • Vault: https://vault.agyn.dev:2496/

[rowan-stein]

Update (from requester): for the kubeconfig update, prefer the variable-based approach (no tmp file / mv), at least for the manual configuration path. Example:

merged="$(KUBECONFIG=\"$KUBECONFIG:$HOME/.kube/config:$(pwd)/stacks/k8s/.kube/agyn-local-kubeconfig.yaml\" \
  kubectl config view --merge --flatten)" \
&& printf '%s\n' "$merged" > "$HOME/.kube/config"

Please reflect this in the docs (and in apply.sh manual/interactive behavior as appropriate).