create-cascade-skill

Install CascadeTUI SKILL.md files for external coding agents.

Overview

create-cascade-skill installs CascadeTUI skills for external coding agents using an interactive CLI. It detects supported agents on your machine and can install for one or many in a single run.

npx create-cascade-skill

The installer writes global SKILL.md files (with Agent Skills frontmatter) into the appropriate agent-specific directories.

Supported agents

Supported agents include:

Codex, Claude Code, Cursor, Windsurf, Cline, Roo Code, Continue, Aider, Gemini CLI, OpenHands, Goose, Ami, Factory.

Run with --list to see which agents are supported and which are detected on your machine.

npx create-cascade-skill --list

Mode	What it does	When to use
`interactive`	Prompts you to select which agents to install for.	Best default; avoids remembering agent IDs.
`--agents`	Installs for the specified agent IDs.	CI / scripting / repeatable setup.
`--all-detected`	Installs for all agents detected on the machine.	Fast setup across multiple editors/agents.
`--dry-run`	Shows what would be written without writing files.	Auditing changes before installing.

CLI options

Options:
  -a, --agents <ids>    Comma-separated agent IDs to install
  --all-detected        Install for all detected agents
  --list                Print supported and detected agents
  --dry-run             Preview files without writing
  -h, --help            Show help

# Install for all detected agents
npx create-cascade-skill --all-detected

# Install for a specific agent
npx create-cascade-skill --agents cursor

# Preview what would be written
npx create-cascade-skill --agents cursor --dry-run

Tip: the supported agent IDs are lowercase (e.g. claude-code). The UI labels may be displayed with spacing/case for readability.

What gets installed

The installer writes one or more SKILL.md files into agent-specific global skill directories (depending on the agent). These files contain “Agent Skills” frontmatter and instructions that help the agent use CascadeTUI effectively.

If you maintain your own customized skills, run with --dry-run first and compare what would change.