Omni Skills
Provides a unified skill store and instruction synchronization for JetBrains Junie and JetBrains AI Assistant, enabling access to personal skill repositories and shared instruction files via symlinks and MCP.
Provides a unified skill store and instruction synchronization for OpenAI Codex, enabling access to personal skill repositories and shared instruction files via MCP tools like list_skills and read_skill.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Omni Skillslist all my skills"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Omni Skills

I try every AI tool. My skills scattered everywhere. This unifies them.
๐ v1.5.0 is out โ automatic pre-sync backups, session-aware consent, a smaller core CLI (
sync,setup,doctor,restore,update,help), and secondary commands moved undermanage. Read the changelog โ
A CLI + MCP server that takes the skills you've collected across all your AI coding tools and unifies them into a single canonical store. Then it syncs them into every tool you use โ Claude, Codex, Kimi, Gemini, Cursor, Zed, Cline, Z.AI, and anything else that follows.
If you find this useful, please consider
starring the repo โ it helps
us reach the criteria for inclusion in homebrew/core so users can run
brew install omni-skills without the brew tap step.
The Problem This Solves
Every new AI coding tool wants its own instruction file, skill directory, and configuration. The result is fragmentation:
Mess | What I Found |
Duplicate AGENTS.md files | 70 copies of the same template across 16 projects |
Orphaned skills | 15 skills hidden in a tool's skills directory, invisible to other tools |
Marketplace dumps | 203 generic skills cloned into a project |
Stale templates | Old project-only rules in every repo |
Private skills leaking | Project-specific skills mixed with generic skills |
Iron laws drifting | A 98KB instruction file manually copy-pasted between repos |
New tool = manual setup | Every new AI tool meant copying config files one by one |
Omni Skills is a single-purpose tool: find all your skills, put them in one place, and wire them into every AI tool you use.
# One canonical store. Every tool sees it.
omni-skills sync all # Claude, Codex, Kimi, Gemini, Cursor, Cline, Zed, etc.Related MCP server: skillsmp-mcp-server
Install
Choose one of the following. Native binaries do not require Node.js.
# macOS / Linux
brew tap moatazhamada/tap
brew install omni-skills
# Windows
winget install omni-skills
# npm
npm install -g ai-omni-skillsSee docs/install.md for details on why brew tap is needed
and how each distribution channel works.
Quick Start
You need your own private repository for your skills. This toolkit does not include skills โ it connects the skills you already have to every AI tool you use.
# 1. Run the auto-scan setup
omni-skills setup
# Auto-detects your installed AI tools, finds your skill directories,
# suggests cleanup, and generates ~/.config/skills/config.json
# 2. Sync to every AI tool
omni-skills sync all
# 3. Verify health
omni-skills doctor
# 4. Run the verification suite
node verify.jsOverride defaults without questions:
omni-skills setup --public=~/my-skills-public --private=~/my-skills-private
omni-skills setup --toolkit=/path/to/this/repoWhat It Does
Tier | What | Mechanism |
A |
| Symlinks + local MCP server exposing |
B | Hooks | Canonical hooks transpiled into each tool's native hook format |
C | Other assets | Symlinks from your config repo to per-tool paths |
When you run omni-skills sync, each SKILL.md is compiled on the fly into the
format each target tool expects. You can also add executionBoundary guards in
the YAML frontmatter to restrict file writing, network access, or require shell
execution confirmations per tool.
See docs/architecture.md for the full compiler pipeline, execution boundaries, and MCP server details.
Supported AI Tools
Tool | Type | Config | Skills Dir | MCP |
Claude Code | CLI |
| โ | โ |
OpenAI Codex | CLI |
| โ | โ |
Kimi | CLI |
| โ | โ |
Gemini CLI | CLI |
| โ | โ |
Cursor | Editor |
| โ | โ |
Kilocode | VS Code |
| โ | โ |
OpenCode | CLI |
| โ | โ |
Aider | CLI |
| โ | โ |
Continue.dev | VS Code |
| โ | โ |
Cline | VS Code |
| โ | โ |
Roo Code | VS Code |
| โ | โ |
Windsurf | Editor |
| โ | โ |
Zed | Editor |
| โ | โ |
Tabby | Self-hosted |
| โ | โ |
PearAI | Editor |
| โ | โ |
Void | Editor |
| โ | โ |
JetBrains Junie | IDE |
| โ | โ |
JetBrains AI Assistant | IDE |
| โ | โ |
Claude Desktop | Desktop |
| โ | โ |
Devin | Desktop | โ | โ | โ |
Factory Droid | CLI | โ | โ | โ |
Z.AI (GLM 5.2) | Model API | โ | โ | โยน |
ยน Z.AI works through existing tools. Point Claude Code, Cline, or Zed at the
GLM Coding Plan endpoint. For VS Code Copilot, install the glm-copilot
extension.
CLI Commands
Core commands are the ones you run every day. Secondary commands are grouped
under manage to keep the main surface small.
Core commands
Command | Does |
| Wire instructions, skills dirs, MCP config, hooks, assets. |
| Auto-scan, detect tools, generate config. |
| Health check symlinks, config, indexes, skill counts. |
| List or restore trashed files, prune old trash. |
| Check for updates. |
| Show help. |
Secondary commands (omni-skills manage <subcommand>)
Subcommand | Does |
| Run the MCP server over stdio. |
| Regenerate |
| List or run chainable skill workflows. |
| Find orphaned skills and dangling symlinks. |
| Scan and sort instruction files public/private. |
| Scan system for AI tools, instruction files, skill dirs. |
| Usage statistics and improvement tips. |
| Interactive setup: scan, classify, route, wire. |
| SkillSpector vulnerability scanning. |
| Create a new skill with a wizard. |
| Convert an existing file into a skill. |
| Remove omni-skills wiring. |
Legacy top-level commands (e.g., omni-skills check) still work for backward
compatibility, but new scripts should prefer omni-skills manage check.
See docs/usage.md for detailed usage, real stats, health checks, verification, and tests.
Why These CLI Changes?
The original CLI had ~15 top-level commands. That made the tool look heavier
than it is and increased the chance of running the wrong thing. The refactor
keeps the daily surface tiny (sync, setup, doctor, restore) and moves
everything else behind manage. The goal is less friction, not more
features.
Automatic Pre-Sync Backup
omni-skills sync can rewrite instruction files, MCP configs, hooks, and skill
directories. Before it writes anything, it copies the current versions to:
~/.config/skills/backups/<ISO-timestamp>/If a sync ever breaks a tool config, you can restore manually from that
directory. Backups are skipped with --dry-run (nothing is changed) or
--no-backup (you know what you're doing). Old backups are pruned after 30
days automatically.
Session-Aware Consent
Several commands (sync, setup, check, classify, init) default to a
dry-run preview before they write anything. If you run several of them in a row,
re-confirming every time is tedious; --yes is risky because it persists for
only that single invocation. Use the environment variable instead:
export OMNI_SKILLS_SESSION_YES=1
omni-skills setup
omni-skills sync all
omni-skills doctorThis skips confirmations for the current shell session only. When the shell closes, the override disappears.
Picking the Right AI Tool for the Task
Omni Skills keeps your skills in one place; it does not tell you which AI agent
to use for a given task. For that, see
ai-tool-router โ a separate,
zero-dependency skill that scans your installed agents, remembers their costs
and strengths, and suggests the cheapest capable tool (plus a fallback).
Documentation
License
MIT โ see LICENSE.
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/moatazhamada/ai-omni-skills'
If you have feedback or need assistance with the MCP directory API, please join our Discord server