🧠 Up-Skill For Agent — Master AI Agent Skills Hub

⚡ Quick Start

# Copy desired skill folders into your project workspace
mkdir -p .gemini/config/skills   # For Google Antigravity IDE
# or mkdir -p .claude/skills  # For Claude Code

cp -r ./caveman .gemini/config/skills
cp -r ./cyber-security .gemini/config/skills
cp -r ./fullstack-dev .gemini/config/skills

# Invoke directly in prompt or slash command
> /dev "Implement secure JWT authentication with RS256 and refresh token rotation"
> /ctf kerberoasting
> /caveman "Explain how to fix React memory leak in 2 sentences"

⚡ Command Syntax Reference

🧭 The 3 Pillars

📂 Directory Structure

Up-Skill-For-Agent/
│
├── ⚡ caveman/               # Caveman terse communication mode (saves 65% output tokens)
├── ⚡ caveman-commit/        # Ultra-compressed Conventional Commits generator (≤50 chars)
├── ⚡ caveman-compress/      # Utility for permanent Memory file (.md) token reduction
├── ⚡ caveman-review/        # 1-line per defect actionable Code Review comments
│
├── 🛡️ cyber-security/       # Master Cybersecurity Engineering & CTF Library (817 Skills)
│   ├── dev/                 # 7-Phase SSDLC & 10 Golden Rules of Secure Coding (/dev)
│   └── ctf/                 # Master Router for 817 CTF Playbooks & 1,095 automation scripts (/ctf)
│
└── 🚀 fullstack-dev/         # Enterprise-grade Full-Stack Architecture Guide
    ├── SKILL.md             # Mandatory 5-Step Workflow
    └── references/          # 8 in-depth technical reference manuals (API, DB, Auth, Deploy)

🧩 Core Modules Deep-Dive

1. ⚡ Caveman Suite — Token-Saving Engine

Eliminates unnecessary filler while preserving technical substance, code blocks, errors, and CLI commands verbatim:

• caveman/: Adjusts assistant response verbosity across 4 intensity levels:
  • lite: Removes pleasantries and intros; maintains complete sentences
  • full (default): Drops articles and filler words; saves 65% tokens
  • ultra: Bullet-point facts only; maximum technical density
  • wenyan: Classical Chinese compression mode for absolute minimum character tokens
• caveman-commit/: Generates Conventional Commits git messages prioritizing "Why" over "What"
• caveman-compress/: Compresses natural language memory files (CLAUDE.md, TODOs) by ~46%
• caveman-review/: Emits compact 1-line code review feedback (L<line>: <problem>. <fix>.) with severity badges (🔴 bug, 🟡 risk, 🔵 nit, ❓ q)

2. 🛡️ Cyber Security: DevSecOps (/dev) & CTF Playbooks (/ctf)

Hub covering software development security (SSDLC / DevSecOps) and operational security testing (Red Team / Blue Team / CTF):

🔹 Developer & DevSecOps (cyber-security/dev/SKILL.md — /dev)

• 7-Phase SSDLC Workflow: Threat Modeling (STRIDE) ➔ Secure Architecture ➔ Secure Scaffolding ➔ Secure Coding (10 Rules) ➔ DevSecOps Verification (SAST/SCA/Secret Scan) ➔ Container & Supply Chain (SBOM/Cosign) ➔ Production Hardening & Audit Logging
• 10 Golden Rules of Secure Coding: Concrete Vulnerable ❌ vs Secure ✅ code patterns (SQLi, BOLA/IDOR, XSS/CSP, SSRF, Argon2id, Safe Deserialization, AI Safety & Guardrails, Race Conditions FOR UPDATE, Rate Limiting, RFC 9457 Error Handling)
• Self-Contained Checklists: End-to-end security checklists covering Web/API, Auth/IAM, AI/LLM, and Containers

🔹 CTF & Security Operations (cyber-security/ctf/SKILL.md — /ctf)

• Master Index of 817 Skills: Structured into 16 security domains with clean relative navigation links and MITRE ATT&CK / NIST CSF mappings
• 6-Step Universal CTF Resolution Framework: Systematic methodology from challenge triage and primitive extraction to automated exploit crafting and flag verification
• 1,095 Automation Scripts: Ready-to-run Python scripts/agent.py in each playbook for testing and emulation

3. 🚀 Fullstack Dev — Production-Grade Architecture

Architectural guidelines rooted in Clean Architecture, Domain-Driven Design (DDD), and The Twelve-Factor App methodology:

Mandatory 5-Step Workflow

1. Step 0: Gather Requirements — Identify Tech Stack, Database, API style, Auth mechanism, and Real-time needs
2. Step 1: Architectural Decisions — Design Feature-first folder layout, Error hierarchies, and Caching strategies
3. Step 2: Scaffold with Checklist — Validate codebase boilerplate against an enterprise scaffolding checklist
4. Step 3: Implement Patterns — Build Domain/Service/Repository layers with strict schema validation
5. Step 4: Test & Verify — Validate compilation, health check endpoints (/health), and contract tests
6. Step 5: Handoff Summary — Document run instructions, environment variables, and critical architectures

Technical Reference Guides (references/)

• api-design.md — 50+ rules for REST, GraphQL, and gRPC design
• auth-flow.md — Complete auth architecture (JWT Rotation, OAuth2/OIDC, RBAC)
• db-schema.md — Schema design, indexing strategy, and concurrency locking
• django-best-practices.md — High-performance Django & DRF patterns
• environment-management.md — Configuration and secrets management per 12-Factor
• release-checklist.md — Pre-flight zero-downtime deployment checklist
• technology-selection.md — Tech stack evaluation and trade-off criteria
• testing-strategy.md — Testing Pyramid strategy (Unit, Integration, Contract, E2E)

🛠️ Platform Installation Guide

1. Google Antigravity IDE

• Copy skill folders to .gemini/config/skills

2. Claude Code

• Copy skill folders to .claude/skills/
• Or append to CLAUDE.md: Use caveman mode for concise responses.

3. Cursor & Windsurf

• Integrate rules and guidelines from SKILL.md into .cursorrules (Cursor) or .windsurfrules (Windsurf)

📊 Feature Comparison Matrix

📜 License

This project is open-source software licensed under the MIT License. Feel free to use, modify, and distribute.