Skip to content

Conversation

@jellydn
Copy link

@jellydn jellydn commented Jan 10, 2026

What

This PR introduces an online installer script to simplify the setup of Ralph. It also enables the installer to automatically configure the recommended setting for users.

Why

  • To make it easier and faster for users to start using Ralph.
  • To ensure that the feature is enabled by default, improving the out-of-the-box experience.
  • To allow users to easily install the PRD and Ralph skills globally.

How

Run the installer script:

curl -sSL https://raw.githubusercontent.com/jellydn/ralph/feature/online-installer/install.sh | bash
  1. Verify that the scripts/ralph directory is created with ralph.sh and prompt.md.
  2. Check that scripts/ralph/ralph.sh is executable.
  3. Verify that your ~/.config/amp/settings.json file now includes the amp.experimental.autoHandoff setting.
  4. Install skills globally and verify they are in ~/.config/amp/skills/.

Adds an install.sh script for easy installation with curl. The script also automatically configures the recommended amp.experimental.autoHandoff setting in the user's ~/.config/amp/settings.json file.

Co-Authored-By: Claude <[email protected]>
@jellydn jellydn marked this pull request as ready for review January 10, 2026 13:56
@jellydn jellydn changed the title feat: Add online installer feat: Add online installer and auto-config for autoHandoff Jan 10, 2026
@jellydn jellydn force-pushed the feature/online-installer branch from af64da3 to 45d8cd1 Compare January 10, 2026 14:40
@snarktank
Copy link
Owner

Thanks @jellydn - can you share a video of you using this and verifying it works?

@jellydn
Copy link
Author

jellydn commented Jan 11, 2026

@snarktank Sure, here you are: https://s.itman.fyi/G0bNBgJJ

@snarktank
Copy link
Owner

Hey, thanks so much for putting this together! Really appreciate you taking the time to contribute.

I love the idea of making installation easier, but I have some security concerns with the curl | bash pattern that I'm not comfortable shipping:

  • If the repo were ever compromised, everyone installing would automatically run malicious code
  • No way for users to verify the script before it modifies their system
  • Pulling from main means untested changes could go out immediately

These are industry-wide concerns with this pattern, not anything specific to your implementation.

Would you be open to reworking this to either:

  1. Provide a download-then-run approach with checksums, or
  2. Use versioned releases instead of main?

Totally understand if you don't have time for that - just wanted to explain why I can't merge as-is. Thanks again for trying to improve the repo.

@jellydn
Copy link
Author

jellydn commented Jan 12, 2026

Hi, I see your point and here is the plan Are you okay with this approach? Thanks.

@jellydn
Copy link
Author

jellydn commented Jan 13, 2026

@snarktank
Copy link
Owner

Thanks for putting together this plan and testing it! This looks good to me.

One small request before I merge: can you add a download-verify-run option to the README alongside the curl | bash? Something like:

# Or download and inspect first
curl -O https://github.com/snarktank/ralph/releases/latest/download/install-from-release.sh
less install-from-release.sh  # inspect
bash install-from-release.sh

That way users who want to verify before running have a clear path. Once that's in, I'll merge.

Add installation section to README with options to:
- Quick install from latest release via curl pipe
- Download and inspect script before running for security
- Install specific version by setting VERSION environment variable

This provides users with flexible installation methods and follows security best practices by offering script inspection option.
@jellydn
Copy link
Author

jellydn commented Jan 16, 2026

Sure @snarktank No problem, thanks.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants