HSHSKY Lab

AI Coding Tutorials

Step-by-step guides tested on real code — Claude Code, Cursor, Agent SDK, and more.

7 tutorials

tutorials2026-07-20

Why Claude Code Suddenly Burns Tokens 10x Faster: The Prompt Cache Nobody Talks About (2026)

Usage that jumps 0 to 100% on one prompt is usually a cache miss, not a bug. How Claude Code's prompt cache works and what silently invalidates it.

tutorials2026-07-17

Claude Code Context Management: How to Stop Long Sessions From Getting Slow, Dumb, and Expensive (2026)

Long Claude Code sessions get slow, dumb, and pricey as context fills. How to fight context rot with /clear, /compact, /context, and CLAUDE.md.

tutorials2026-07-16

Can Your Employer See What You Do in Claude Code? What Admins Actually See (2026)

Used a company Claude account for a side project? Exactly what admins can see — your usage vs your prompts vs your repo — plus the IP risk nobody mentions.

tutorials2026-07-13

Claude Effort Levels Explained: Low, Medium, High, xhigh, Max — and When Higher Just Burns Your Limits (2026)

How Claude's effort levels (low/medium/high/xhigh/max) change output, why they differ from the thinking toggle, and when higher effort just burns your limits.

tutorials2026-07-12

Claude Code Usage Limits Explained: 5-Hour Windows, Weekly Caps, and How to Stop Hitting Them (2026)

Two timers, token-based counting, and an extra-usage bill that surprises people. How Claude Code's limits actually work, and how heavy users dodge the wall.

tutorials2026-06-10

Claude Code After One Month: An Honest Daily-Use Review

I ran Claude Code 8+ hours a day for a month and wrote almost zero code by hand. Here's what actually worked, what confused me about token usage, and who I'd recommend it to.

tutorials2026-05-21

How to Use Claude Code Subagents (Complete Guide 2026)

Claude Code subagents let you parallelize research, isolate risky changes in worktrees, and delegate background work — here's exactly how to use them and when they're worth it.