Bamuhigire Peter signature
Français

Public Claude Code & Codex skills engines

Eleven public AI skills engines, installable as Claude Code and Codex plugins

I am Peter Bamuhigire, a technology and business consultant. Each engine below is a public GitHub repository of routed SKILL.md files that installs in one command — as a native Claude Code plugin (/plugin marketplace add) or by a cross-platform script that Codex and other agent runners can use too.

View all repositories on GitHub ↗How the workflow works
Infographic showing Peter Bamuhigire's original ten AI skills engines organised around an AI-assisted delivery loop, ending with a Kaizen continuous-improvement footer
The original ten engines at a glance, closing with the Kaizen loop: review, refine, reuse. Windows Admin Engine Skills joined as the eleventh engine after this diagram was published.

One map, eleven engines

Every engine traces back to one operating loop

Software engineering, design systems, digital research, business plans, social media and marketing, website systems, Linux and Windows infrastructure, accounting doctrine, consulting proposals, and SDLC documentation all sit under the same AI-assisted delivery loop: plan, create, refine, deploy. None of it is treated as finished — Kaizen keeps every engine under continuous review, refinement, and reuse rather than freezing it at its first version.

What is an AI skills engine?

An AI skills engine is a structured catalogue of instructions, decision rules, evidence requirements, and quality gates that helps an agent or practitioner route work consistently. Each one packages as a Claude Code plugin and a Codex-compatible installer script, so I use the engines as the reasoning and control layer around AI tools — not as a magic prompt, an unsupervised replacement for specialists, or a claim that every task can be automated.

Use this directory when you need to

  • • choose the right engine for a project;
  • • install a skills engine into Claude Code or Codex in one command;
  • • inspect the public source before adopting a workflow;
  • • keep a human in the middle of every consequential workflow.

Want several at once? chwezi-engine-agents is a shared marketplace and installer for adding multiple engines as one suite.

My AI practice

The tool is only one part of the system

I use an AI tool for the work it is good at — drafting, coding, classifying, comparing, summarising, or exploring — then use the relevant skills engine to decide what must be checked, what evidence is required, which specialist route applies, and where a person must approve the result. Human review is not an optional final polish; it is the control point in the workflow.

See the complete operating method →

Context

The right engine gives the AI tool the job, constraints, inputs, and boundaries.

Evidence

Sources, tests, calculations, and repository facts keep fluent output from becoming false confidence.

Human in the middle

A person reviews, corrects, approves, escalates, rejects, stops, or rolls back consequential work.

Choose a route

Eleven engines, eleven different jobs

Start with the problem you are trying to solve. Follow the website explanation for the install command and boundaries, then inspect the linked GitHub repository for the actual skills and references.

Linux server room with security and infrastructure symbols representing Linux operations

Safe Linux operations

Linux Skills

Family-aware server operations for Debian/Ubuntu and RHEL systems, including Bash, networking, services, security, storage, backups, and recovery.

48 SKILL.md files · Claude Code + Codex plugin

Read the guide → GitHub ↗

01 · Frame

Name the real work

Define the decision, audience, evidence, constraints, and consequence of getting the work wrong before selecting an engine.

02 · Route

Choose the smallest accurate engine

Use one specialist engine where possible and add companion engines only when the work crosses into finance, visual design, research, engineering, or formal documentation.

03 · Verify

Keep human authority visible

Check sources, review outputs, record uncertainty, test failure paths, and keep approval, release, and rollback decisions with accountable people.

Ready to discuss your project?

Every engagement begins with a conversation. Book a consultation to explore how Peter's experience can serve your organisation.

Book a Consultation