@ramarivera/pi-skill-selector

Pi extension that lets you type $ to fuzzy-pick skills and insert /skill:name into the prompt.

Packages

Package details

extension

Install @ramarivera/pi-skill-selector from npm and Pi will load the resources declared by the package manifest.

$ pi install npm:@ramarivera/pi-skill-selector
Package
@ramarivera/pi-skill-selector
Version
0.1.7
Published
May 9, 2026
Downloads
884/mo · 101/wk
Author
ramarivera
License
MIT
Types
extension
Size
17.1 KB
Dependencies
0 dependencies · 2 peers
Pi manifest JSON
{
  "extensions": [
    "./src/index.ts"
  ]
}

Security note

Pi packages can execute code and influence agent behavior. Review the source before installing third-party packages.

README

@ramarivera/pi-skill-selector

Pi extension that turns $ into a fuzzy skill selector.

What it does

  • listens to raw terminal input so it works alongside editor-owning extensions like pi-powerline-footer
  • intercepts $ anywhere in the prompt before the active editor consumes it
  • opens an overlay skill picker
  • fuzzy-filters skills case-insensitively by name and description
  • inserts /skill:<name> at the cursor so Pi's built-in skill expansion loads the selected skill
  • also provides /skill-selector as a command fallback

Install

From npm

pi install npm:@ramarivera/pi-skill-selector

From GitHub

pi install git:github.com/ramarivera/pi-skill-selector

From a local checkout

pi install /absolute/path/to/pi-skill-selector

Use

In interactive Pi, type:

$

Pick a skill, press Enter, and the extension inserts:

/skill:<skill-name> 

You can also run:

/skill-selector

Development

bun install
bun test
bun run check

For local Pi auto-discovery while developing, this repo includes a .pi/extensions/pi-skill-selector/index.ts shim that re-exports src/index.ts.

Publishing

Publishing is handled by .github/workflows/publish.yml using npm trusted publishing. Before the workflow can publish, configure the package on npm with ramarivera/pi-skill-selector and workflow .github/workflows/publish.yml as a trusted publisher.

Notes

This package is meant to be installed as a Pi package, so it declares its extension entry under the pi key in package.json.

License

MIT