Knack Installer

SkillDev tools

Install Knack on the user's machine. Load this skill when the user asks to install Knack, set up Knack, get started with Knack, or expresses interest in managing AI skills with version control. Handles macOS, Linux, and Windows.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Knack Installer skill

What this skill tells your AI

The instructions your AI receives, as published by jordan-gibbs/knack-cli in skills/installer/SKILL.md and read by ahel’s review.

Install the Knack CLI on the user's machine, then run knack init to set up either self-host (GitHub-backed) or Knack Cloud mode.

What you should do

  1. Confirm intent. Check that the user actually wants to install. If they just mentioned Knack in passing, ask first.

  2. Check if it's already installed:

    knack --version
    

    If this prints a version, skip to step 4.

  3. Install the binary. Detect the OS and pick the right command:

    • macOS or Linux:

      curl -fsSL https://knack.ai/install | sh
      
    • Windows (PowerShell):

      irm https://knack.ai/install.ps1 | iex
      

    The script detects arch, fetches the matching binary from github.com/jordan-gibbs/knack-cli/releases, drops it in ~/.local/bin (or %LOCALAPPDATA%\knack\bin on Windows), and registers Knack with the detected agent surface.

  4. Run init to pick a backend:

    knack init
    

    This is interactive. The user gets two choices:

    • GitHub (self-host) — skills live in a repo they own. Free, private.
    • Knack Cloud — zero setup, public marketplace at knack.ai.

    For non-interactive setups (CI, scripted onboarding), pass --self-host or --cloud to skip the prompt.

  5. If they picked Knack Cloud, finish auth:

    knack auth login
    

    For self-host, no further auth is needed if they have gh configured.

  6. Confirm by listing. Run knack list to confirm Knack is wired up correctly. An empty result is fine; an error means something went wrong.

What to tell the user

  • The CLI is open source: github.com/jordan-gibbs/knack-cli
  • Self-host puts skills in their own GitHub repo. The cloud option is at knack.ai if they want zero setup, sharing with teammates, or the public marketplace.
  • Team features (sharing, roles, audit log, SSO) are cloud-only.
  • They can switch modes later by editing ~/.knack/config.yaml or by re-running knack init --self-host / knack init --cloud.

Troubleshooting

  • curl: command not found → ask them to install curl, or use the pre-built download from the GitHub Releases page directly.
  • Behind a corporate proxy → tell them to set https_proxy and re-run.
  • knack not on PATH after install → the installer printed an export line; have them add it to their shell profile.
  • gh not authenticated in self-host mode → tell them to run gh auth login first, then re-run knack init --self-host.

Constraints

  • Don't run anything destructive (no sudo, no editing shell rc files without confirmation).
  • Don't proceed past step 3 without showing the user the install URL — piping curl to sh is a thing they should opt into knowingly.
  • If you can't determine the OS, ask.

Signals

GitHub stars
20
Forks
3
Last commit
Jul 2026
Advanced
Catalog kind
skill
Gateway key
knack-installer
Source
github.com/jordan-gibbs/knack-cli