Skip to content
Founding Lifetime: every Pro skill, pay onceFirst 100 members · $99 onceUpgrade
Lookbook
Sign In

Install a skill in any agent

Install a Lookbook skill in Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI or OpenCode: where each looks, how to check it loaded and how to run it.

For
Anyone installing a skill in a specific agent
Written for
Claude CodeCodexCursorGitHub CopilotGemini CLIOpenCodeAny agent that reads SKILL.md folders
Time
5 min read · updated Oct 26, 2026

The short answer

  1. Run curl -fsSL https://lookbk.dev/cli.js | node - add <skill>, or paste the skill’s install prompt into your agent.
  2. Approve the sign-in link it shows, then start a new agent session.
  3. Check the agent lists the skill, and ask for the job in plain words.

What you need

  • Node.js 20 or newer, for the installer.
  • A Lookbook account. Every install signs in, free skills included.
  • Whatever the skill itself needs, such as Chrome or FFmpeg. Its SKILL.md lists these under “What you need”.

The install command

  1. Step 1: Run the command

    Replace <skill> with the last part of the skill page’s address, such as kinetic-title-drop. It downloads the installer and runs it once. Nothing is added to your PATH.

    Terminal
    curl -fsSL https://lookbk.dev/cli.js | node - add <skill>
  2. Step 2: Approve the sign-in

    The first time, it prints a link and a short code. Open the link, check the page shows the same code, and choose Approve. In your own terminal the command waits and carries on. In an agent’s shell it prints the link and exits with code 3, so run the same command again after you approve.

  3. Step 3: Start a new session

    The installer prints one Installed … → line for each folder it wrote. Start a new agent session so it picks up the skill.

You can also copy the install prompt from the skill page and paste it into your agent. The agent runs the same command and shows you the sign-in link. For cloud agents, run the command from your repository root with --project, then commit the folders it writes.

Which folders the installer writes
Where you run itFolders written
Inside Claude Code~/.claude/skills/<skill> only
Your own terminal, with Claude Code installed~/.agents/skills/<skill> and ~/.claude/skills/<skill>
Your own terminal, without Claude Code~/.agents/skills/<skill> only
Any of these with --projectThe same folders, inside the current folder
--claude or --sharedOnly the Claude Code folder, or only ~/.agents/skills

Install by hand

Download the ZIP from the skill page while signed in. It holds one folder named after the skill, so unzip it straight into a skills folder and SKILL.md ends up one level down. Keep every file, including NOTICE.md and the hidden .lookbook folder.

Terminal
unzip ~/Downloads/<skill>-<version>.zip -d ~/.agents/skills
ls ~/.agents/skills/<skill>/SKILL.md

Your agent

  • Looks in: ~/.claude/skills for every project, and .claude/skills in the folder you start in and each parent up to the repository root. Its docs don’t list .agents/skills.
  • Check: type /skills, or ask “What skills are available?”. Claude Code watches its skills folders. If ~/.claude/skills is new, run /reload-skills.
  • Run: ask in plain words, or start a message with /<skill>.

Common questions

Which folder do I use if I switch agents?
The installer’s default covers all six. Cursor, OpenCode and VS Code read both ~/.agents/skills and ~/.claude/skills; if one lists a skill twice, delete one copy.
Why did the install stop after printing a link?
In an agent’s shell, the command can’t wait for you. Approve the link in your browser, then run the same command again.
Do cloud agents see my installed skills?
No. They start from a copy of your repository, so install with --project and commit the folder. Read the terms before you commit a skill to a repository other people can read.

Sources

What this guide says about other companies’ tools comes from their own documentation, checked on Oct 26, 2026.

Did this guide help?

See a skill’s result before you install it.

Every skill page shows what it makes, what it needs and which agents can run it.