Skip to content

Skills Overview

Josemalyson Oliveira edited this page Jun 27, 2026 · 1 revision

Skills Overview

Custom Skills includes 5 production-ready skills for AI coding assistants. Each skill is a self-contained module with structured workflows, references, and templates.


Skills at a Glance

Skill Purpose Time Output
Agent-Code-Detective Reverse-engineer AI agent codebases 3-5h Investigation dossier
HTML-First Generate self-contained HTML artifacts 5-30min Single-file HTML
K8s-Investigation K8s/VPS security & performance analysis 2-4h Interactive HTML report
Learning-Trail Validated learning paths 1-2h Markdown learning trail
Pre-Commit-Quality-Gate Auto-generate pre-commit pipelines 5-15min Bash script

Skill Architecture

Each skill follows a consistent structure:

skill-name/
├── SKILL.md                    # Main definition (YAML frontmatter + instructions)
├── references/                 # Supporting documents (loaded on-demand)
│   ├── reference-1.md
│   └── reference-2.md
└── templates/                  # Ready-to-use starting points (optional)
    └── template.html

SKILL.md Format

---
name: skill-name
description: >
  Trigger phrases that activate this skill.
  Be specific about when it should activate.
---

# Skill Name

## Overview
What this skill does.

## Workflow
Phase-by-phase instructions.

## References
Links to supporting documents.

Design Principles

1. Fail-Fast Ordering

Cheap checks first, expensive last. The pre-commit pipeline runs secret scanning before mutation testing.

2. Trust by Inspection

Validate resources by reading them, not by reputation. The learning-trail skill verifies every URL before recommending.

3. Follow the Data

Read code by tracing execution flow, not top-to-bottom. The code-detective uses breakpoints, not line-by-line reading.

4. Single-File, Self-Contained

No external dependencies required for output artifacts. HTML templates work offline.

5. Soft-Skip Over Hard-Fail

Infrastructure-dependent steps warn instead of blocking. The pre-commit pipeline gracefully handles missing tools.

6. Atomic Transactions

Each recommendation either passes all validation steps or is rejected. No partial approvals.


Interoperability

Skills can reference and enhance each other:

  • k8s-investigation output uses html-first format
  • pre-commit-quality-gate can be triggered by other skills that need commit-time checks
  • learning-trail can generate trails for any skill's domain

Creating New Skills

See Creating-Custom-Skills for how to add your own skills to the collection.


Last updated: June 2026

Clone this wiki locally