APEX OS Bad Boy

“Everything you need. Nothing you don’t. Mastery at your fingertips.”

Overview

APEX OS Bad Boy is a comprehensive productivity system for AI agents that provides:

• 45+ integrated skills across all domains
• Intelligent skill recommendations (75%+ accuracy)
• Risk-based planning levels (SAFE/MID/RISKY)
• Systematic workflows for every situation
• Best-in-class practices from top engineering teams

This skill teaches agents WHEN and HOW to use the right tools at the right time.

The Core Philosophy

1. The Planning Slider

SAFE MODE ──────────────────────────────────▶ RISKY MODE

Full Design    Quick Plan           Just Do It
   ↓               ↓                    ↓
Plan → Code   Plan + Code         Code → Fix
   ↓               ↓                    ↓
Test → Ship   Review → Ship       Ship → Fix

2. The 75% Rule

If a skill has 75%+ probability of helping, invoke it. Always.

3. Right Tool, Right Job

Never use a hammer when you need a scalpel. Know your tools.

The Risk Level Selector

Before ANY task, answer these 3 questions:

Q1: What’s the impact if this breaks?

• [A] Production / Users affected → SAFE MODE
• [B] Dev environment only → MID MODE
• [C] Easy to revert / POC → RISKY MODE

Q2: How complex is this?

• [A] Multi-component / New integration → SAFE MODE
• [B] Single feature / Bug fix → MID MODE
• [C] Simple / Well-understood → RISKY MODE

Q3: How well do we understand the problem?

• [A] New territory / Unclear requirements → SAFE MODE
• [B] Known domain, some unknowns → MID MODE
• [C] Clear and straightforward → RISKY MODE

Scoring:

• Mostly A → SAFE MODE (80% of tasks)
• Mostly B → MID MODE (15% of tasks)
• Mostly C → RISKY MODE (5% of tasks)

Skill Registry

META PROCESS (How you work)

FRONTEND

BACKEND

QUALITY

SPECIALTY

Task → Skill Decision Matrix

Universal Decision Tree

"I have an idea / want to build something new"
  → brainstorming (MUST)
    → writing-plans → [domain skill]

"I need to implement multiple tasks"
  → subagent-driven-development

"Something is broken / not working"
  → systematic-debugging → [relevant domain skill]

"I'm writing code and want to make sure it's good"
  → receiving-code-review

"I'm about to claim something is done"
  → verification-before-completion (MUST)

Domain-Specific

Workflows

SAFE MODE Workflow

brainstorming → writing-plans → tdd-master → implementation → 
verification → code review → deploy

Required Skills:

• ✅ brainstorming (must)
• ✅ writing-plans (must)
• ✅ tdd-master (should)
• ✅ verification-before-completion (must)
• ✅ receiving-code-review (must)

MID MODE Workflow

quick plan → implementation → review → ship

Required Skills:

• ✅ writing-plans (brief)
• ✅ systematic-debugging (if bugs)
• ✅ verification-before-completion (must)
• ✅ receiving-code-review (quick)

RISKY MODE Workflow

JUST DO IT → Ship Fast → Fix Fast

⚠️ ONLY for: POCs, experiments, prototypes

Required Skills:

• ✅ verification-before-completion (quick check)
• ✅ systematic-debugging (if breaks)

Anti-Patterns (NEVER DO)

Process Anti-Patterns

• ✗ Never skip brainstorming for “simple” tasks
• ✗ Never claim done without verification
• ✗ Never skip code review
• ✗ Never use RISKY mode for production code
• ✗ Never skip security review for auth/payment

Code Anti-Patterns

• ✗ Never commit secrets to git
• ✗ Never use eval() with user input
• ✗ Never skip RLS on Supabase
• ✗ Never write empty catch blocks
• ✗ Never use console.log in production

Skills.sh Integration

One-Command Setup

Install all Vercel agent skills with a single command:

npx skills add vercel-labs/agent-skills

This installs 5 premium skills (100+ rules total):

Auto-Update Workflow

To check for and install updates:

npx skills add vercel-labs/agent-skills --force

Process:

1. Run the command to fetch latest version
2. Report what will be updated (new skills, changes)
3. ASK USER for confirmation before applying
4. Execute update only after human approval

Browse More Skills

Visit https://skills.sh to discover additional skills from the community.

Required Skills Location

~/.agents/skills/           # Domain skills (from skills.sh)
~/.config/opencode/skills/  # Meta skills
~/.opencode/skills/         # Coordination skills

Quick Reference

┌─────────────────────────────────────────────────────────────┐
│  ALWAYS INVOKE                                          │
│  • brainstorming for new features                        │
│  • verification-before-completion before claiming done   │
│  • security-best-practices for auth/data                │
│  • receiving-code-review before merge                   │
├─────────────────────────────────────────────────────────────┤
│  HIGH PRIORITY                                           │
│  • writing-plans after brainstorming                     │
│  • subagent-driven-development for multi-task           │
│  • tdd-master for test-driven work                      │
│  • vercel-react-best-practices for React/Next.js       │
│  • supabase-expert for database                         │
├─────────────────────────────────────────────────────────────┤
│  AS NEEDED                                               │
│  • browser-use for automation                           │
│  • web-design-guidelines for UI                         │
│  • systematic-debugging for bugs                        │
│  • performance-optimization for speed                   │
├─────────────────────────────────────────────────────────────┤
│  SKILLS.SH (run these commands)                          │
│  npx skills add vercel-labs/agent-skills     # Install  │
│  npx skills add vercel-labs/agent-skills --force  # Update│
└─────────────────────────────────────────────────────────────┘

APEX OS Bad Boy v8.0 – “Everything you need. Nothing you don’t.”