Photo of DeepakNess DeepakNess

Unified way to manage Agent Skills

Unproofread notes

Implementation of Agent Skills by different AI tools is getting out of hand now, because each tool has their own separate paths for the skills folder. For instance, take a look at these leading AI coding assistants:

Tool Name Directory Path
Amp .agents/skills/<skill-name>/
Antigravity .agent/skills/<skill-name>/
Claude Code .claude/skills/<skill-name>/
Codex .codex/skills/<skill-name>/
GitHub Copilot .github/skills/<skill-name>/
Cursor .cursor/skills/<skill-name>/
OpenCode .opencode/skill/<skill-name>/

What's the solution to this madness, then?

One sane solution is this dotagents TUI tool by Ian Nuttall that basically acts as one location for all of your hooks, commands, skills, and AGENT/CLAUDE.md files. I also found this tool called agentctl that claims to provide better management of different agents.

I also discovered this discussion about a standard folder structure for managing agent skills for different AI tools.

Update:

Came across another interesting tool rulesync (created by dyoshikawa1993) that manages this in a slightly better way. It's a CLI tool that automatically generates configuration files for various AI development tools from unified AI rule files.

RuleSync is much more polished and maintained as it supports rules, ignore, mcp, commands, subagents, and skills for multiple AI assistants like Claude Code, Codex CLI, GitHub Copilot, Cursor, OpenCode, Google Antigravity, and much more.

Comment via email