rxmos-skills

Production-focused skill pack for engineering agents.

This repository contains reusable skills designed to make AI-assisted software work safer, more verifiable, and less generic.

Why this repository exists

Most AI outputs fail in production for predictable reasons: vague plans, missing verification, no rollout control, and weak rollback guidance.

rxmos-skills addresses that by encoding practical guardrails directly into skills:

• evidence-driven planning over generic prose,
• phase-by-phase delivery with explicit Go/No-Go gates,
• measurable verification commands,
• rollout and rollback safety as first-class requirements.

Available skills

refactor-plan-pro

Builds a repository-grounded refactor plan with:

• dependency-ordered phases,
• compatibility and migration strategy,
• quality gates with pass/fail criteria,
• test, observability, and non-functional validation,
• rollout strategy and rollback runbook.

Best for refactors that touch architecture, shared contracts, core flows, reliability, or performance.

Installation

Install directly from GitHub:

npx skills add yurirxmos/rxmos-skills --skill refactor-plan-pro

Install from a local clone:

npx skills add . --skill refactor-plan-pro

List available local skills:

npx skills add . --list

Usage

After installing the skill, invoke it from your agent workflow with a clear refactor goal and repository context.

Recommended prompt shape:

Create an execution-ready refactor plan for [scope].
Include phased gates, verification commands, rollout strategy, and rollback runbook.
Ground all claims in repository files/contracts/tests.

What "good" output looks like

A high-quality plan generated by this skill should:

• reference real files and modules,
• include concrete verification commands in every phase,
• define measurable gate criteria,
• stop progression on failures with corrective actions,
• include explicit rollout and rollback procedures.

Repository structure

skills/
  refactor-plan-pro/
    SKILL.md

Contributing

Contributions are welcome, especially improvements that increase:

• execution safety,
• determinism and testability,
• compatibility handling during migrations,
• observability and rollback quality.

When proposing changes, prefer practical examples and verifiable constraints over stylistic wording.