KKKKhazix/khazix-skills

What it solves

This project provides a collection of structured instruction sets, called "Skills," designed to extend the capabilities of AI agents (such as Claude Code, Codex, and Cursor). It solves the problem of repetitive prompting and the lack of of standardized, high-quality instructions for complex tasks like goal definition, system cleanup, research, and specialized writing.

How it works

Each "Skill" is a structured instruction set following the Agent Skills open standard. These can be loaded directly by compatible agents or manually pasted into a conversation as project rules. The collection includes several specialized tools:

• leader: Converts vague ideas into precise, actionable goal documents (task books) that allow agents to work independently for long periods.
• storage-analyzer: Scans local disks (Mac/Windows) and generates an interactive HTML report with categorized cleanup suggestions (Green/Yellow/Red) for the user to approve.
• aihot: Fetches daily AI news and trends from aihot.virxact.com without requiring an API key.
• neat-freak: Synchronizes project documentation, agent memory, and rules (e.g., CLAUDE.md) after a task is completed to prevent "knowledge rot."
• hv-analysis: Performs deep research on products, companies, or concepts using a longitudinal (historical) and lateral (competitive) analysis method, producing a detailed PDF report.
• khazix-writer: Applies a specific, opinionated writing style and a four-layer self-check system to generate long-form articles.

Who it’s for

Users of AI agents (like Claude Code, Codex, Kimi Code, etc.) who want to automate complex workflows, maintain clean project documentation, or perform deep research and specialized writing using standardized instructions.

Highlights

• Standardized: Follows the Agent Skills open standard for easy installation across 40+ compatible agents.
• Goal-Oriented: The leader skill focuses on defining "what not to do" (Harness) to prevent agents from taking inefficient shortcuts.
• Safe Disk Cleanup: The storage-analyzer uses a read-only scan and requires explicit browser-based confirmation for any deletions.
• Knowledge Maintenance: The neat-freak skill ensures that project rules and documentation stay aligned with the actual code state.
• Deep Research: The hv-analysis skill combines historical evolution and competitive landscape analysis into a long-form PDF report.