5 Claude Code Skills You Can Steal

Open-source skills with full walkthroughs

What This Is

Skills are the thing I wish someone had shown me in month one of using Claude Code. Instead of typing the same context into every session, you write it once in a markdown file and invoke it with a single command. Claude reads the file, loads the instructions, and gets to work.

This guide walks through five skills I use daily — all published on GitHub, all free to copy. For each one you'll see the actual skill file structure, what problem it solves, how to install it, and where to change things to fit your own setup.

  • Morning brief — Calendar, tasks, email, and project health in one command.
  • Log to daily — Session summaries appended to Obsidian automatically.
  • Draft first — Route mechanical writing to a local model, refine with Claude.
  • Think first — Mental model enforcement before any significant decision.
  • Lessons learned — Structured retrospectives that encode fixes into the system.

None of this is theoretical. These are skills running in my Claude Code setup right now, invoked across real sessions. The walkthroughs show the actual skill files — not sanitised examples.

Time commitment: About 45 minutes to read all six chapters. Another 30 minutes if you install as you go.

  1. What Skills Are and Why They Matter

    Skills are markdown files that give Claude persistent context and instructions. One command loads them. Here's what they are, why they beat copy-pasting prompts, and how to install one in about three minutes.

  2. Morning Brief: Your Daily Dashboard

    One command pulls your calendar, tasks, email, newsletter pipeline, and project health into a single briefing. Walk through the skill file section by section, and see how to customise it for your own tools.

  3. Log to Daily: Never Lose Context

    Every session produces something worth keeping. This skill appends a structured summary to your Obsidian daily note — completed work, decisions made, files touched, and next steps. The append pattern explained.

  4. Draft First: Save Money With Local LLMs

    The first draft of anything mechanical doesn't need Claude. This skill routes initial generation to a local model, then brings Claude in for quality control. How the two-stage pattern works and what it costs to ignore it.

  5. Think First: Stop Reacting, Start Reasoning

    Claude's default is to answer fast. This skill intercepts decisions before implementation, applies mental models from first principles to inversion, and forces structured analysis before any code gets written.

  6. Lessons Learned: Build Memory Into Your System

    After an incident, most people write it up and forget it. This skill runs a structured retrospective — timeline, root cause, contributing factors — and encodes the fix directly into the system so the same mistake can't happen twice.