Claude Code Plugins

Community-maintained marketplace

Feedback

Use when starting ANY task. This skill ensures you check for and use relevant skills before proceeding. MANDATORY for all requests.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name using-skills
description Use when starting ANY task. This skill ensures you check for and use relevant skills before proceeding. MANDATORY for all requests.

Using CodeAssist Skills

Core Principle

Before tackling any task, you MUST identify and execute relevant skills. This is mandatory, not optional.

Mandatory First Response Protocol

Complete this checklist for EVERY request:

  1. Inventory: Mentally list all available skills in .claude/skills/
  2. Assess: Determine if ANY skill applies to this work
  3. Read: If applicable, use the Read tool to load the skill file
  4. Announce: Publicly state which skill you're deploying
    • Format: "I'm using [Skill Name] to [purpose]"
  5. Execute: Follow the skill exactly as documented

Critical Enforcement Rules

Skills are non-negotiable. If a skill exists for your task, using it is MANDATORY.

The Enforcement Problem

Skills usage "gets lost" over time - agents stop using skills after extended sessions or across multiple projects.

This is UNACCEPTABLE.

Solution: Regular Skill Reminders

Every 10 tasks OR every hour, whichever comes first:

⚠️ SKILLS FRAMEWORK CHECK ⚠️

Before continuing, verify:
1. ✅ Am I using skills for ALL tasks?
2. ✅ Did I read skill files (not just remember)?
3. ✅ Am I announcing skill usage?
4. ✅ Am I following skill protocols exactly?

If NO to any: STOP and recommit to skills framework.

Last skill used: [name]
Tasks since last reminder: [count]

This reminder is MANDATORY and automatic - do NOT skip it.

Skill Usage Must Be Consistent Across Projects

Problem: AI uses skills differently in different projects.

Solution: Skills framework is UNIVERSAL.

  • ✅ Same skill files in ALL projects (.claude/skills/)
  • ✅ Same enforcement in ALL projects
  • ✅ Same protocols in ALL projects
  • ✅ NO project-specific variations

If skills are being used differently across projects, you are doing it WRONG.

Common Rationalizations to Reject

  • ❌ "This is straightforward" → Still check for skills
  • ❌ "I recall this skill already" → Read the current file anyway
  • ❌ "The skill seems excessive" → Use it regardless
  • ❌ "Let me gather information first" → Skills define how to gather information
  • ❌ "I'll just do it quickly" → Speed comes FROM following skills
  • ❌ "This is a small change" → Small changes need skills too
  • ❌ "I'm under time pressure" → Skills SAVE time and prevent errors
  • ❌ "I used skills in the last project" → Use them EVERY project
  • ❌ "I'm just making a small commit" → verification-before-completion is MANDATORY

Why Skills Are Mandatory

  1. Authority: Skills represent proven, tested approaches
  2. Consistency: Same pattern across all projects
  3. Safety: Skills include critical safety checks
  4. Learning: Skills compound knowledge over time
  5. Reliability: Skills prevent known errors

Checklist-Based Skills

When a skill contains a checklist:

  • Create individual TodoWrite todos for EACH item
  • Never work through checklists mentally
  • Never batch multiple steps into single tasks
  • Mark each todo as completed individually

Transparency Requirement

Always announce skill usage plainly:

"I'm using [Skill Name] to [purpose]."

This:

  • ✅ Clarifies your reasoning
  • ✅ Allows user to catch errors early
  • ✅ Documents which skills are being used
  • ✅ Reinforces skill-first mindset
  • ✅ Builds trust through transparency

Red Flags

  • ❌ Never start implementation without checking skills
  • ❌ Never rationalize skipping a relevant skill
  • ❌ Never assume you remember a skill's current content
  • ❌ Never work on "autopilot" without skill guidance
  • ❌ Never batch multiple skills mentally

Skill Categories

Core Workflow

  • brainstorming - Design before implementation
  • writing-plans - Break work into tasks
  • executing-plans - Execute with verification
  • code-review - Review before completing

Framework Setup

  • laravel-api-setup - Laravel API-first projects
  • nextjs-pwa-setup - PWA with Next.js
  • react-native-setup - Native mobile apps
  • python-fastapi-setup - Python API projects

Safety (CRITICAL)

  • database-backup - MANDATORY before database operations
  • pre-commit-hooks - Code quality enforcement
  • verification-before-completion - Final checks

Testing

  • test-driven-development - RED/GREEN/REFACTOR
  • paratest-setup - Parallel testing for PHP
  • testing-anti-patterns - What NOT to do

Workflow

  • git-workflow - Branching and commits
  • git-worktrees - Parallel feature development
  • dispatching-parallel-agents - Multi-task workflow

Meta

  • writing-skills - Creating new skills
  • testing-skills - Validating skills with pressure tests

Skill Discovery by Task Type

"I'm starting a new Laravel project"

brainstorminglaravel-api-setup

"I need to run migrations/tests"

database-backup ⚠️ CRITICAL (ALWAYS)

"I'm adding a new feature"

brainstormingwriting-planstest-driven-development

"I need to work on multiple features"

git-worktrees

"I'm ready to commit my changes"

git-workflow

"I'm finishing a feature"

code-reviewverification-before-completion

"I need to create a new skill"

writing-skillstesting-skills

Pressure Testing Awareness

Skills are designed to work ESPECIALLY when you're under pressure. Common pressure situations where you MUST still use skills:

  • Time Pressure: "It's 5 PM Friday"
  • 💰 Sunk Cost: "I've been working on this for 3 days"
  • 👔 Authority: "My manager is waiting"
  • 🔥 Urgency: "Production is down"
  • 💪 Confidence: "I know what I'm doing"

In ALL these situations: Check and use skills anyway.


Bottom Line: Skills represent accumulated best practices. Bypassing them means repeating solved problems and recreating known errors. The few seconds it takes to check skills saves hours of debugging later.