agent-scripts: a centralized hub for managing shared AI agent rules and reusable workflow skills across local workspaces

What it solves

It provides a centralized system for managing shared instructions, reusable skills, and portable helper scripts for AI agents (specifically Codex and Claude Code) across multiple local workspaces. This prevents the need to duplicate rules and workflows across every individual project repository.

How it works

The project uses a canonical repository to store global agent rules (AGENTS.MD) and a library of skills (defined in SKILL.md files with YAML front matter). A synchronization script (sync-skills) creates symlinks in the agent's configuration directories (e.g., ~/.codex/ and ~/.claude/), allowing different AI agents to discover and use these shared skills and rules regardless of which project the user is currently working in.

Who it’s for

Developers who use AI agents like Claude Code or Codex across many different repositories and want a consistent set of "hard rules" and operational workflows (skills) that follow them across their entire local environment.

Highlights

• Centralized Rule Management: Uses a single AGENTS.MD file that is symlinked to multiple agent configurations to ensure consistent behavior.
• Skill Routing: Implements a structured skill system with YAML metadata to help agents trigger specific workflows via short phrases.
• Cross-Repo Integration: Supports symlinking skills from other repositories into a central hub, allowing repo-owned skills to remain canonical in their own projects while being globally available.
• Portable Helpers: Includes dependency-light scripts for tasks like skill validation, commit management, and Chrome DevTools automation.