agentleFS
bjw-s-labsbjw-s-labs/home-ops/.agents/skills/renovate-review/SKILL.md

renovate-review

Skill869 starsChanged 33 days agoApache-2.0View on GitHub

Review Renovate dependency updates in this Flux/GitOps home-ops repository, including Helm OCI charts, container tags/digests, Talos/Kubernetes, GitHub Actions, Terraform/OpenTofu, Docker Compose, and Renovate presets. Investigate rendered impact, upstream changes and actual repository dependencies before recommending a merge.

1 other project uses a version of this, 1 of them edited.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 1

    1. joryirving/home-ops★ 23247% same

    Wrote their own version · 0

    None found.

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository3 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    CI action / Renovate preset / IaC: Read the upstream release/compare, then verify this repository's .forgejo/workflows/ inputs/outputs, …

    line 36
  2. Distinctive:

    Follow referenced upstream issues/PRs when a changelog only says "See #123".

    line 37
  3. Very specific:

    Check raised minimum versions, removed APIs or fields, changed auth/storage/network defaults, one-way migrations, and security advisories …

    line 39
  4. Distinctive:

    Search the actual consumers of anything removed or renamed, not just whether the new value is explicitly set:

    line 43
  5. Distinctive:

    No fabricated Konflate results, upstream comparisons, tool calls or post-merge checks.

    line 58

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Review a Renovate PR in home-ops

Determine what changes on deployment and whether this repository relies on the old behavior. Flux can reconcile kubernetes/ changes soon after merge. This skill is loaded by the AI PR Review workflow for renovate/* branches alongside .agents/instructions/pr-review.instructions.md; it supplements the general review procedure and bundled reviewer. The repository's trusted AGENTS.md standards remain authoritative.

1. Establish the update and deployment footprint

  • Identify every dependency and old/new version, digest or chart tag. Renovate groups may contain several independent upgrades. Applications live under kubernetes/apps/<namespace>/<app>/; shared components live under kubernetes/components/. Trace all consumers of a changed component before judging impact.
  • Distinguish OCIRepository chart tag updates, container image/digest updates, HelmRelease values changes, Talos/Kubernetes upgrades, GitHub Actions, OpenTofu providers/modules, Docker Compose and Renovate configuration. An OCI chart is pinned by ref.tag rather than a container-style @sha256: digest. Do not flag that as missing image pinning.
  • The actual rendered resources, not the one-line git diff, determine most Helm/Flux compatibility risks.

2. Inspect the rendered change (when applicable)

The reviewer is configured with the Konflate evidence provider and read-only MCP tools, rather than the upstream example repository's flate setup. Prefer evidence already included in the review corpus:

  • mcp__konflate__get_pr_summary with this PR number: affected resources and clusters, additions/removals, image changes, data-loss, immutable field, RBAC, suspend/prune, and render errors.
  • mcp__konflate__get_pr_diff with this PR number when chart defaults, templates, selectors or Helm values could change rendered manifests while the git diff looks trivial.
  • mcp__konflate__list_pull_requests only if necessary to establish whether this PR has render evidence. If Konflate did not render it, say unverified, never "unchanged" or "safe".
  • For a local manual investigation use flate diff all --path ./kubernetes/flux/cluster --base origin/main; do not claim to have run this command unless an independently allowed tool actually did. A SKILL.md does not grant shell or Kubernetes execution rights.

Pay special attention to changes in resource names, label values as well as keys, selectors, ports, services, RBAC, storage classes, CRDs, probes, immutable fields and persistent-volume behavior. Konflate may omit CRDs/Secret data; their absence from a render is not evidence of no change. One unrelated rendering failure is not proof that the upgraded chart fails.

For a digest-only container refresh with unchanged repository/tag, a rendered digest change plus consistent upstream metadata is usually enough: don't invent Kubernetes breaking changes.

3. Verify upstream changes

Begin with the PR's Release Notes, but inspect upstream when absent, truncated, a major/0.x-minor bump, a jump across multiple versions, or a chart/image with material runtime changes. For Helm charts and images, AGENTS.md requires real upstream evidence via available read-only tools, not just the Renovate description.

  • OCI chart: Find its source in the app's ocirepository.yaml, follow the publisher's chart releases/templates and compare versions. For mirrored ghcr.io/home-operations/charts-mirror/<chart>, use home-operations/charts-mirror metadata to locate the real chart source. When the chart's appVersion changes, inspect the underlying application's release notes too.
  • Container image: Resolve upstream from the image owner or org.opencontainers.image.source metadata. For ghcr.io/home-operations/*, also inspect home-operations/containers build definitions/history for digest-only changes. Check image revision/source/created metadata when available.
  • Talos / Kubernetes: Check kubernetes/talos/version.yaml, the installer template in kubernetes/talos/cluster.yaml.j2, node overrides under kubernetes/talos/nodes/, and the upgrade recipes in kubernetes/talos/mod.just. Verify the specific Talos release supports the target Kubernetes version against the published Sidero/Talos compatibility matrix before concluding compatibility. Cite the matrix; a patch bump alone is not proof.
  • CI action / Renovate preset / IaC: Read the upstream release/compare, then verify this repository's .forgejo/workflows/ inputs/outputs, .renovate/ rules, ansible/ automation, or docker/ runtime configuration as applicable. Do not assume a dependency's generic deployment layout exists here.
  • Follow referenced upstream issues/PRs when a changelog only says "See #123". Don't infer changed defaults or a migration solely from a release-note headline.

Check raised minimum versions, removed APIs or fields, changed auth/storage/network defaults, one-way migrations, and security advisories relevant to the deployed configuration. If an upstream source cannot be verified, label the evidence gap honestly.

4. Search for repository-specific exposure

Search the actual consumers of anything removed or renamed, not just whether the new value is explicitly set:

  • Under kubernetes/, inspect selectors and labels in Prometheus/ServiceMonitors, alerts and dashboards, NetworkPolicies, HTTPRoutes/Gateway policies, PDBs, Flux dependsOn/healthChecks, values/valuesFrom/postRenderers, service DNS and CRD instances.
  • Trace application Kustomizations and shared-component usage from kubernetes/flux/cluster/. Cilium, Envoy Gateway, external-secrets, Rook/Ceph, PostgreSQL and storage-provider updates merit close inspection for compatibility and data safety where affected.
  • For app-template or shared chart changes, look for multiple affected HelmReleases. Chart defaults can change even when values in this repo do not.
  • In ansible/, docker/, .forgejo/ and .renovate/, search the corresponding consumers rather than assuming everything is a Kubernetes release.

A documented home-ops convention is not a finding. Apply the namespace and image-pinning policies in AGENTS.md; do not substitute generic linting heuristics for this repository's deployment conventions.

5. Verdict and evidence

  • Respect AGENTS.md and the reviewer's deterministic must_check requirements. Do not replace or weaken review gates. A breaking change with no affected consumer in this repository is a brief observation, not an invented blocker.
  • A finding must identify the changed line, the affected consumer/path, the concrete failure mode, and supporting rendered or upstream evidence. Confirm a replacement name/value from upstream manifests rather than guessing an auto-fix.
  • Escalate documented risks: unsupported Talos/Kubernetes pairings, CRD/schema incompatibilities, data-loss/storage mutations, new security relaxations, unexpected suspend/prune, and resource/selector renames that break live consumers. For uncertain evidence, report Unknown / needs verification instead of asserting a failure or successful test.
  • For an ordinary same-tag digest refresh, keep the review compact: recommendation, what changed, and any meaningful caveat. Skip empty Standards Compliance, Linked Issue Fit, Tool Harness or evidence sections, and don't dump internal planner diagnostics.
  • No fabricated Konflate results, upstream comparisons, tool calls or post-merge checks. A skill describes how to review; only configured tools may be used, and nothing here authorizes shell commands or writes.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.