xingfanxia

AX Skills — Development skill for Claude Code

Development community

Production-grade Claude Code skills by AX.

How to install AX Skills

This entry records only its repository, not the path inside it, so there is no exact command to give. Open xingfanxia/AX-skills and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What AX Skills does

Production-grade Claude Code skills by AX. First skill: jewelry-marketing — one photo in → full XHS marketing bundle out for jewelry e-commerce merchants.

Alternatives in Development

Gemini CLI Foundations — This repository is a comprehensive skills library for Gemini CLI - reusable, production-ready skill packages t 5.3k ★
IPolloWork — Enterprise-grade, local-first Agent Workbench for people and agent teams 4.9k ★
Internet Court Skill — The trust layer for agent-to-agent commerce — natural-language mandates, ERC-7710 delegated permissions, x402 4.9k ★

README

AX-skills

A collection of production-grade Claude Code skills built by [@xingfanxia](https://github.com/xingfanxia) (AX). Each skill is self-contained, runs on Claude Code / OpenAI Codex / OpenClaw, and ships with examples + tests.

**What's a Skill?** A reusable slash-command-style capability for AI coding agents — bundles a `SKILL.md` (how to use), Python entrypoints, prompt templates, and reference docs. The agent invokes it like `/jewelry-marketing` and the skill takes over.

Featured — what actually gets used

Ordering below is backed by real invocation counts from the author's own sessions (as of 2026-07, ~500 sessions), plus the skills the author reaches for when it matters. The full catalog is further down.

Skill	Why it's here
`autonomous-grind`	The single most-invoked skill in the author's setup (182 real invocations) — the discipline layer that stops autonomous `/goal` loops from wrapping up early with a pretty status report. SKILL contract + 5 enforcement hooks.
`big-task`	The risk-tiered engineering workflow router (54 invocations, 23 commits of evolution) — routes every non-trivial task to direct / light / full-cycle process by risk, not file count, and runs it autonomously with an auditable decision log.
`tui-engineering`	Build phone-friendly terminal UIs with zero terminal libraries — architecture, ANSI math, mouse/touch gestures, narrow-mode, headless tests. Comes with a runnable reference repo.
`ai-workflow-spine`	Design AI-native workflows with explicit deterministic/AI boundaries, replayable artifacts, trigger/no-op/stop contracts, and offline-first verification.
`deep-research`	Multi-agent verification research with source-authority tiers — the skill to reach for before a real decision.
`apple-pdf`	The default "make this