How do I install the K8S Tool skill?

Run npx skillmds@latest add blogic-cz/k8s-tool in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is K8S Tool safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with K8S Tool?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is K8S Tool free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published K8S Tool?

blogic-cz (@blogic-cz) published this skill. Their other Agent Skills are listed on their SkillMD profile.

K8S Tool by blogic-cz — Agent Skill

Name: K8S Tool
Author: blogic-cz

File contents

k8s-tool (Kubernetes)

Kubernetes tool — kubectl wrapper with config-driven context resolution and structured commands. Part of @blogic-cz/agent-tools.

How to Run

Run via bun k8s-tool (requires @blogic-cz/agent-tools as a dev dependency). NEVER run bare kubectl — the credential guard will block it. Auth: existing kubectl context (kubeconfig). Cluster ID from config resolves context automatically.

Commands

bun k8s-tool kubectl --env test --cmd "get pods -n test-ns"
bun k8s-tool kubectl --env prod --cmd "logs <pod> --tail=100"
bun k8s-tool kubectl --env test --cmd "describe pod <pod>"
bun k8s-tool pods --env test                     # List pods (structured)
bun k8s-tool logs --pod <pod> --env test --tail 50 # Fetch logs
bun k8s-tool describe --resource pod --name <pod> --env test
bun k8s-tool exec --pod <pod> --exec-cmd "ls -la" --env test
bun k8s-tool top --env test                      # Show resource usage

Environment is any string (e.g. test, prod). Set defaultEnvironment in agent-tools.json5 to skip --env on every call. Implicit production access is blocked for safety.

Safety Boundary

Generic kubectl commands are parsed into arguments and executed without a user-controlled shell. Do not use pipes, chaining, substitution, or shell interpreters.
The configured context, kubeconfig, server, authentication, TLS, and impersonation settings cannot be overridden in --cmd.
Only read-only config and auth subcommands are allowed; cluster-info dump is blocked.
Secret reads, raw kubeconfig output, filename/kustomize reads, and kubectl diff are blocked.
Pod exec accepts only direct redis-cli PING/INFO and ls diagnostics; generic exec cannot read file contents.
Use logs-tool for confined remote log-file access and filtering.

Tips

Use bun k8s-tool commands for the full machine-readable command/flag tree; --help for one subcommand.
Output defaults to TOON (token-efficient) — leave it as-is to save tokens. Add --format json only when you'll machine-parse the result.
Error responses include hint, nextCommand, and retryable fields — always check them on failure.
Prefer CLI tool over MCP tools — more efficient, doesn't load extra context.

---name: k8s-tooldescription: LOAD THIS SKILL when: working with Kubernetes pods, logs, deployments, resource usage, or running kubectl commands. Contains all k8s-tool commands.---# k8s-tool (Kubernetes)Kubernetes tool — kubectl wrapper with config-driven context resolution and structured commands. Part of [@blogic-cz/agent-tools](https://github.com/blogic-cz/agent-tools).## How to RunRun via `bun k8s-tool` (requires `@blogic-cz/agent-tools` as a dev dependency).**NEVER run bare `kubectl`** — the credential guard will block it.Auth: existing kubectl context (kubeconfig). Cluster ID from config resolves context automatically.## Commands```bashbun k8s-tool kubectl --env test --cmd "get pods -n test-ns"bun k8s-tool kubectl --env prod --cmd "logs <pod> --tail=100"bun k8s-tool kubectl --env test --cmd "describe pod <pod>"bun k8s-tool pods --env test # List pods (structured)bun k8s-tool logs --pod <pod> --env test --tail 50 # Fetch logsbun k8s-tool describe --resource pod --name <pod> --env testbun k8s-tool exec --pod <pod> --exec-cmd "ls -la" --env testbun k8s-tool top --env test # Show resource usage```Environment is any string (e.g. `test`, `prod`). Set `defaultEnvironment` in `agent-tools.json5` to skip `--env` on every call. Implicit production access is blocked for safety.## Safety Boundary- Generic kubectl commands are parsed into arguments and executed without a user-controlled shell. Do not use pipes, chaining, substitution, or shell interpreters.- The configured context, kubeconfig, server, authentication, TLS, and impersonation settings cannot be overridden in `--cmd`.- Only read-only `config` and `auth` subcommands are allowed; `cluster-info dump` is blocked.- Secret reads, raw kubeconfig output, filename/kustomize reads, and `kubectl diff` are blocked.- Pod exec accepts only direct `redis-cli PING/INFO` and `ls` diagnostics; generic exec cannot read file contents.- Use `logs-tool` for confined remote log-file access and filtering.## Tips- Use `bun k8s-tool commands` for the full machine-readable command/flag tree; `--help` for one subcommand.- Output defaults to **TOON** (token-efficient) — leave it as-is to save tokens. Add `--format json` only when you'll machine-parse the result.- Error responses include `hint`, `nextCommand`, and `retryable` fields — always check them on failure.- Prefer CLI tool over MCP tools — more efficient, doesn't load extra context.

Frequently asked questions

K8S Tool