helmadm 0.1.2

Generate Argo CD Application manifests from Helm releases in Kubernetes

helmadm

CLI tools for Helm 3 releases stored in Kubernetes: generate Argo CD Application YAML, export reproducible Helm install bundles, list releases, and compare a release manifest to live cluster objects. No helm or kubectl binary required for helmadm itself.

Requirements

• Python 3.12+
• uv
• Access to a Kubernetes cluster (kubeconfig or in-cluster)

Install

uv sync

Commands

Command	Purpose
argocd-yaml	Print an Argo CD Application manifest for a release (stdout)
pull	Export a reproducible Helm install bundle (values files + README)
ls	List Helm releases in the cluster
drift	Compare the release's stored manifest to live objects (read-only)

uv run helmadm --help
uv run helmadm argocd-yaml --help
uv run helmadm pull --help
uv run helmadm ls --help
uv run helmadm drift --help

argocd-yaml — Application manifest

Reads the release from cluster storage, diffs coalesced chart values + user config against helm show values for the chart version (fetched from the chart repo), and writes overrides to spec.source.helm.valuesObject. Non-helm fields use CHANGE_ME placeholders.

uv run helmadm ls -n monitoring
uv run helmadm argocd-yaml -n monitoring prometheus

If the release has no chart.metadata.repoURL, pass the repository URL (see NEEDS_REPO_URL in ls):

uv run helmadm argocd-yaml -n monitoring prometheus \
  --repo-url https://prometheus-community.github.io/helm-charts

--debug adds a .debug block to the YAML (raw values, diff metadata, ignoreAnnotations). Remove it before applying to Argo CD.

pull — reproducible install bundle

Reads a release from cluster storage and writes a directory you can use with plain helm install / helm upgrade (no helmadm or Argo required for reinstall).

uv run helmadm pull -n loki -o ./bundles fluentbit

Creates ./bundles/{namespace}/{release}/ containing:

• helmadm-pull-metadata.yaml — when pulled, kubeconfig/context, release and chart info
• {chart}.all.values.yaml — effective values from the cluster (coalesced)
• {chart}.changed.values.yaml — overrides only (diff vs remote chart defaults)
• {chart}.remote-all.values.yaml — chart defaults from the repository
• README.md — helm install/template commands for each values file

Options:

• --revision N — pull a specific Helm release revision (default: latest)
• --repo-url — when the release has no chart.metadata.repoURL (see NEEDS_REPO_URL in ls)
• --repo-name — helm repo alias used in the README (default: derived from the repo URL host)
• --tar — write a gzip tarball of the bundle to stdout instead of leaving files on disk
• --force — overwrite an existing bundle directory

uv run helmadm pull -n monitoring prometheus --revision 3 -o ./bundles
uv run helmadm pull -n loki -o ./bundles fluentbit --tar > fluentbit-bundle.tar.gz

ls — list releases

uv run helmadm ls                  # all namespaces, detailed (default)
uv run helmadm ls -n monitoring    # one namespace
uv run helmadm ls --no-detail      # name / revision / status only

drift — manifest vs live

uv run helmadm drift -n monitoring prometheus
uv run helmadm drift --detect-extras -n monitoring prometheus
uv run helmadm drift -ia -n kube-system traefik   # print normalization notes before each diff

Exit 1 on drift, missing objects, fetch errors, or extras (with --detect-extras).

Environment variables

CLI flags take precedence over env vars.

Flag	Environment variable
-n / --namespace	HELMADM_NAMESPACE (or current kubeconfig context namespace)
--context	HELMADM_CONTEXT
--repo-url	HELMADM_REPO_URL
release name (positional)	HELMADM_RELEASE_NAME
(values trace, with -v)	HELMADM_TRACE_VALUES — per-key logs during argocd-yaml
(Kubernetes HTTP)	HELMADM_K8S_CONNECT_TIMEOUT — connect timeout in seconds (default: 5)
(Kubernetes HTTP)	HELMADM_K8S_READ_TIMEOUT — read timeout in seconds (default: 60)

--kubeconfig follows kubectl: use the flag, or KUBECONFIG / ~/.kube/config.

Development

uv run pytest