k8s-kaniko-job — Claude Code skill: build and install
Builds a container image in a Kubernetes Kaniko Job and pushes it to GHCR. It fits managed deploys running without a Docker daemon.
npx skills add labring/sealos-skills --skill k8s-kaniko-job -a claude-codeWhat k8s-kaniko-job does
The skill runs scripts/kaniko-build.py to package the project context, place it in the local VersityGW store, create a Kaniko Job in the current namespace, and publish the image to GHCR. The command returns an image digest and reference, so deployments can pin the digest instead of relying on a mutable tag.
The build runs inside Kubernetes as linux/amd64, with kubectl, python3, tar, curl, and bash available to the script. GHCR access needs GITHUB_TOKEN with write:packages; the image owner must match the token login after lowercasing. A Dockerfile must be present in the build context.
Use this skill for SEALAI_DEPLOY_MODE=managed. Local interactive deploys use docker buildx, so this workflow is the wrong path for them.
What it helps with
- build an image without a Docker daemon
- send project sources to a Kaniko Job
- push the result to GHCR
- pin a deployment to an image digest
- diagnose a failed context upload
Example prompts for the agent
- Build the project with
scripts/kaniko-build.pyand publish it asghcr.io/<owner>/<repo>:<tag>. - Run a managed build from the project workspace using
.as the context andDockerfileas the Dockerfile, then returnimage_refanddigest. - Inspect the Kaniko Job failure and report redacted findings without printing tokens or Secret payloads.
How to install k8s-kaniko-job
1Skills CLI (Vercel)
npx skills add labring/sealos-skills --skill k8s-kaniko-job -a claude-code2Manually: put the files in place
~/.claude/skills/k8s-kaniko-job/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/k8s-kaniko-job/
1Skills CLI (Vercel)
npx skills add labring/sealos-skills --skill k8s-kaniko-job -a codex2Manually: put the files in place
.agents/skills/k8s-kaniko-job/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add labring/sealos-skills --skill k8s-kaniko-job -a cursor2Manually: put the files in place
.cursor/skills/k8s-kaniko-job/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install labring/sealos-skills k8s-kaniko-job2Manually: put the files in place
.github/skills/k8s-kaniko-job/SKILL.md1Skills CLI (Vercel)
npx skills add labring/sealos-skills --skill k8s-kaniko-job -a gemini-cli2Platform CLI
gemini skills install https://github.com/labring/sealos-skills --path plugins/sealos/skills/k8s-kaniko-jobManually: put the files in place
.devin/skills/k8s-kaniko-job/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add labring/sealos-skills --skill k8s-kaniko-job -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about k8s-kaniko-job
How do I install k8s-kaniko-job in Claude Code?
Run npx skills add labring/sealos-skills --skill k8s-kaniko-job -a claude-code, then check ~/.claude/skills/k8s-kaniko-job/SKILL.md.
What does k8s-kaniko-job do?
It creates a Kaniko Job in Kubernetes, builds an image from project sources, and pushes it to GHCR. The result includes the published image reference and digest.
What does the build require?
You need a sandbox namespace reachable through kubectl, permissions for Jobs, Secrets, Pods, and logs, a GITHUB_TOKEN with write:packages, and a Dockerfile in the context.
