skills.rest
ins13014778
ins13014778COMMUNITY
Explore all skills by this author →
GitHub

xiaocao-clone

Clone Xiao Cao's documentation-first workflow for code, debugging, and refactoring.

3|Updated Apr 12, 2026
Search Tags:#workflow#MCP#code delivery#spec-first#privacy guardrails#bug escalation#user memory
One-click install
npx skills add https://github.com/ins13014778/xiaocao-clone --skill xiaocao-clone
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: xiaocao-clone
Source: https://github.com/ins13014778/xiaocao-clone/tree/main
Command: npx skills add https://github.com/ins13014778/xiaocao-clone --skill xiaocao-clone

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) and assets (resource) components.

What problem does it solve?

You need an AI engineering partner that can reproduce Xiao Cao’s documentation-first workflow, enforce practical safety rules (backups, planning, verification), and build accurate long-term preferences only after your confirmation.

Core Features & Use Cases

  • Documentation-first engineering: read the whole project, discuss architecture and interfaces, write a spec, implement one function at a time, then validate and deliver a per-feature Markdown artifact.
  • Bug-fix escalation guardrails: after repeated failures, switch from blind patching to investigation using primary references and clear evidence.
  • Candidate-memory with confirmation: extract candidate preferences, present a structured summary, and only write into local long-term profile after you explicitly approve, with privacy minimization.

Quick Start

Use xiaocao-clone to take over my project, back up first, write a project understanding doc and a feature spec, then implement one function with tests and a delivery Markdown.

Frequently Asked Questions about xiaocao-clone

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I enforce a spec-first engineering workflow for new feature development?▼

Spec-first engineering workflow is enforced by reading the entire project, discussing architecture, writing a spec, implementing one function at a time, and validating to deliver a per-feature Markdown artifact.

What is the best way to stop an AI from blind patching during bug triage?▼

Bug-fix escalation guardrails stop blind patching by switching to evidence-based investigation using primary references after repeated failures, ensuring clear evidence drives the debugging process.

Can I use this workflow with Claude Code and Codex for platform-adapted code delivery?▼

Yes, platform adapters support Claude Code and Codex for platform-adapted usage, applying workflow rules, MCP ecosystem tools, and plugins across embedded work, UI/browser validation, and deployment ops.

How does user memory distillation work with privacy minimization in AI coding assistants?▼

User memory distillation extracts candidate preferences, presents a structured summary, and only writes to a local long-term profile after explicit confirmation, enforcing privacy minimization.

Do I need to back up my project before starting documentation-first code delivery?▼

Yes, backup rules require backing up the project first before loading the persona, writing a project understanding doc, and generating a feature spec for safe documentation-first code delivery.

What are the limitations of confirmation-gated local user memory for refactoring tasks?▼

Confirmation-gated local user memory limitations include requiring explicit approval before writing preferences and enforcing privacy minimization, meaning no automatic background learning occurs during refactoring.

Related Skills

Explore More Software Engineering Skills→
toilahuonggtoilahuongg
10

readme

Update README.md with documentation from codebase analysis.

Community
Basic
fascarifascari

testing-implementation

Create and run unit and integration tests after feature implementation.

Community
Intermediate
aegnticaegntic
3

multi-platform-architect

Generate platform-adapted AI CLI components for Claude Code, Codex, Auggie, Kilocode, Goose, and Gemini-CLI.

Community
Advanced
aberrantCodeaberrantCode
1

agentic-development

Design and deploy autonomous AI agents with Python or Claude SDK.

Community
Advanced