Skip to content
Back to skills

Usage Hygiene

ASecurity

Practices that keep Claude Code sessions cheap and focused: scope, context management, model choice, forked commands, compaction, when to start a new session. Apply when the user asks how to use Claude efficiently or a session grows long.

  • 8 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 19, 2026
ai-agentsgodebugging

Works with

  • claude code

Security analysis

A100/100

Scanned September 19, 2026

npx -y skills add Adilmunawar/ZD-claude-plugin --skill usage-hygiene --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Usage Hygiene?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Usage Hygiene
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/adilmunawar-usage-hygiene/badge)](https://www.skillsdirectory.com/skills/adilmunawar-usage-hygiene)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: usage-hygiene
description: Practices that keep Claude Code sessions cheap and focused: scope, context management, model choice, forked commands, compaction, when to start a new session. Apply when the user asks how to use Claude efficiently or a session grows long.
---

- Start each task with a precise scope (files, outcome, done criteria); long exploratory sessions cost the most. Use `stack-analyst` or `db-analyst` (read-only) for orientation instead of asking the main session to read everything.
- Keep the working set small: `/context` shows what is loaded; install only the modules a repository needs; disable unused plugins per project.
- Use forked commands (`tech-debt`, `dependency-audit`, `audit`, `study-dashboard`) and agents for anything that produces long output — the main session receives a summary.
- Model by task: `haiku` for validation and formatting, the default model for engineering, the strongest model only for design decisions and hard debugging. Agents inherit the session model unless pinned.
- `/compact` when the conversation exceeds what the current task needs; start a new session after a `/zd-core:handoff` rather than carrying a day's history.
- Watch `/usage` for plan windows and `/zd-usage:report` for where tokens go; set `/zd-usage:budget` per week.
- Avoid pasting large files or logs into chat; give paths and let tools read the parts needed.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…