Claude Code runs shell commands constantly — finding files, searching code, processing JSON, running scripts. The default Unix tools (find, grep, sed, node) work, but modern replacements written in Rust and Zig are significantly faster.

Installing these tools and telling Claude Code to prefer them can noticeably speed up your workflow.

The bare /compact command compresses your conversation but reads everything first—tool output, file contents, web fetches, search results. That takes minutes and often loses important context.

Create ~/.claude/commands/compact-guide.md with one line:

/compact Focus on: current task, file paths modified, decisions made, user preferences stated. Drop: tool output, search results, web content, file contents already acted on.

The npm cowsay package supports loading custom .cow files from any path, not just the built-in ones. This lets you create modified versions with configurable features like dynamic eyes.

Looking at the cowsay source code in lib/cows.js:

1. What is Context Engineering?
2. The Context Engineering Framework
   • 1. In-Code Context
   • 2. Project-Level Instructions
   • 3. Issue-Driven Development

• 4. Interactive Commands

Builds on concepts from Introduction to Markdown.

Markdown files serve a dual purpose when working with AI coding assistants: they're human-readable documentation AND machine-parseable instructions. This tutorial explores the design considerations most users overlook.

When you write a README.md, you think about human readers. When you write a .github/copilot-instructions.md, you're writing for an AI that will interpret your instructions. But here's what many users miss: the AI reads your markdown differently than humans do.

User (2026-01-20 14:30Z)

I need to know how to write markdown

• DOCTYPE: HTML5 with lang="en"
• Head Elements Order:
  1. Meta charset and viewport

2. Title with consistent pattern: "HAP's Color Learning Lab - [Station Name/Description]"

from your JSON audit