imraywang/rayskills

What it solves

rayskills provides a set of reliable, repeatable workflows for AI agents like Claude Code and Codex. Instead of relying on generic prompts, it converts complex, error-prone tasks—such as content production, knowledge management, and platform publishing—into structured, verifiable skills that can be executed, interrupted, and resumed safely.

How it works

At the core is the /ray command, which acts as a router that reads your current context to suggest the next step or execute a full production pipeline. You can use individual skills like /ray-writer for content creation, /ray-cover for design, or /ray-wechat for publishing. Each skill includes built-in safety checks, such as verifying permissions before publishing or ensuring existing notes aren't overwritten, and supports resuming tasks if they are interrupted.

Who it’s for

This tool is designed for individuals or small teams who use AI agents to manage professional workflows—including content creation, knowledge management, and operations—and want to move beyond simple chat-based interactions into reliable, automated production pipelines.

Highlights

• Structured Pipelines: Automates end-to-end tasks like turning an idea into a published article or a video script.
• Safety-First Design: Enforces strict boundaries for high-impact actions like deleting files, publishing content, or modifying production environments.
• Interruptible Workflows: Allows you to pause and resume tasks without losing progress or creating duplicate drafts.
• Verifiable Results: Each skill includes built-in checks to ensure outputs meet quality standards before moving to the next stage.
• Extensible Architecture: Built as a modular toolkit that can be installed via npm and integrated into agent environments like WorkBuddy.