Cli Power Skills

Agentic CLI tool skills for Claude Code. 7 domain-grouped skills covering 26 CLI tools that extend Claude's capabilities beyond its built-in tools.

Installation

claude plugins install cli-power-skills

First add the marketplace, then install:

claude plugins marketplace add ykotik/cli-power-skills
claude plugins install cli-power-skills

Prerequisites

These skills assume the CLI tools are already installed. See INSTALL.md for per-skill and batch install commands (macOS/Homebrew).

Skills

Use Cases

Analyze and transform data without writing scripts

Query a 2GB CSV with SQL, convert YAML configs to JSON, flatten deeply nested API responses into greppable lines, compute column statistics, or join two datasets — all in single commands that pipe together.

Tools: DuckDB for SQL on files, jq for JSON, yq for YAML/TOML/XML, xsv for CSV slicing and stats, miller for format-aware record transforms, gron for grepping nested JSON paths

CSV/JSON/YAML/Parquet ──> jq / yq / xsv / mlr ──> DuckDB SQL ──> export

Secure your code before it ships

Scan dependencies for CVEs, check container images for vulnerabilities, lint shell scripts for quoting bugs and unsafe patterns, validate IaC for misconfigurations, and keep secrets encrypted in version control.

Tools: Trivy for filesystem/container/IaC scanning, ShellCheck for shell script analysis, sops for encrypting secrets in YAML/JSON

code/containers/IaC ──> Trivy scan ──> ShellCheck lint ──> sops encrypt secrets

Test APIs declaratively, not imperatively

Write readable .hurl test files with chained requests, variable captures, and built-in assertions — then probe hundreds of URLs for status and metadata in seconds.

Tools: Hurl for declarative HTTP test suites, httpx for concurrent URL probing

auth ──> create resource ──> verify ──> assert status + body (all in one .hurl file)

Scrape and crawl the web at any scale

Choose the right tool for the job: full browser automation for JS-rendered pages, headless crawling for link discovery, or production-grade frameworks for large-scale extraction.

Tools: Playwright / Puppeteer for browser automation, Scrapy for structured crawling, Crawlee for anti-bot resilience, Katana for fast URL discovery

Katana (discover URLs) ──> Playwright (render JS) ──> Scrapy (extract at scale)

Research the web from your terminal

Extract clean article text from any URL and download video/audio from 1000+ sites with full metadata, format selection, and subtitle support.

Tools: newspaper4k for article extraction, yt-dlp for media download and metadata

newspaper4k (extract text) ──> DuckDB (analyze)
yt-dlp (dump metadata) ──> jq / DuckDB (query)

Manage Python projects at Rust speed

Create virtual environments, install packages, run scripts with inline dependencies, lint and format code — all 10-100x faster than traditional Python tooling.

Tools: uv for package management and venvs, Ruff for linting and formatting

uv venv ──> uv add deps ──> ruff check --fix ──> ruff format ──> uv run pytest

Automate your CI pipeline locally

Run GitHub Actions in Docker before pushing, define project tasks as recipes, generate changelogs from commits, create releases, and back up before destructive operations.

Tools: act for local GH Actions, just for task recipes, git-cliff for changelogs, gh for GitHub API, restic for encrypted backups

act -j test ──> git-cliff changelog ──> gh release create ──> restic backup

Speed and Efficiency vs. Standard Tools

Each skill replaces slower, less capable, or more verbose standard alternatives. The table below shows what each tool replaces and the measured performance advantage.