AI tooling CLI

Share AI bundles without Git noise

skul lets teams reuse AI bundles across Claude Code, Cursor, OpenCode, Codex, GitHub Copilot, and Kiro.

Keep shared prompts, skills, agents, and root instructions in one bundle repo, apply them per project, and leave tool-local files out of Git.

Install Quick Start

Share one bundle repo across Claude Code, Cursor, OpenCode, Codex, GitHub Copilot, and Kiro
Install a specific skill, agent, command, or root instruction from a bundle
Keep AI setup local instead of committing tool-specific dotdirs

Typical session

Go from shared bundle to working project fast

Start with one bundle repo, apply it where needed, and keep project-level AI setup consistent without copy-pasting folders.

skul add github.com/sjquant/ai-bundles core --agent codex

Pull a shared bundle into a project for the tool you use.

skul add core --agent codex --include skills/diagnose

Install one bundle item when you only need a specific skill.

skul apply

Re-apply the same bundle setup in linked worktrees without manual copying.

skul update

Refresh bundles when your team ships new prompts, skills, or agents.

Whole bundle or one item

Install an entire bundle, or pick only the skill, agent, command, or root instruction you need.

One source, many tools

Ship the same AI setup to Claude Code, Cursor, OpenCode, Codex, GitHub Copilot, and Kiro from one source repo.

No repo noise

Keep local AI files out of Git while still giving every project the setup it needs.

Roll out updates fast

Refresh shared prompts, skills, and agents without copy-pasting folders across repos.

Supported AI tools

One bundle model across six coding tools

Reuse the same bundle source across the tools your team already works in instead of rebuilding prompt folders for each one.

Claude Code

Share Claude Code skills, commands, and agents across repos without committing them.

Cursor

Keep Cursor project setup consistent without maintaining a second copy of the same prompts.

Codex

Ship Codex skills and agents from the same source bundle you already use elsewhere.

OpenCode

Support OpenCode projects without inventing a separate packaging workflow.

GitHub Copilot

Share Copilot skills, agents, and instructions while keeping repository policy files controlled.

Kiro

Materialize Kiro skills, agents, and root instructions from the same reusable bundle source.

How it helps

Made for shared AI workflows that stay maintainable

`skul` is useful when AI setup needs to be shared across repos, tools, and teammates without turning every project into a prompt graveyard.

Publish once

Keep reusable prompts, skills, and agents in one repo instead of scattering copies across projects.

Apply per project

Materialize the files each tool expects, or narrow installation to selected bundle items.

Update without cleanup drama

Re-apply or remove managed files predictably when shared AI setup changes.

Docs

Start with what gets you to value fastest

Install first, try one bundle, then keep the command reference nearby when you are ready to wire skul into daily work.

Installation

Install the CLI and get your machine ready in a few minutes.

Install skul

Quick Start

Apply your first shared AI bundle and target the tools you actually use.

Read quick start

Commands

Keep the full command surface nearby once skul becomes part of your daily workflow.

Browse commands