glif-interfaces

a growing list of TypeScript interfaces for Glifs intended to be used in AI agents

Nota bene: these "interfaces" often contain additional functionality that goes beyond what the underlying Glif does. Typical example: many "interfaces" detect the language of the original request - and if that is not english, they first translate the request into english, then run the underlying Glif and finally translate the answer back into the original language (if that translation makes sense).

Important: in order to run Glifs via their API, you will have to sign up for a Glif account and request your personal API Token

Warning: due to a nasty bug in Glif, backticks cannot be entered into the prompt input fields of "Text Generation" blocks, thus damaging my prompts - as a consequence, the code and my examples may not work as expected!

if you like, directly jump to some interactive Usage Examples

Overview

(t.b.w)

Limitations

In order to remain cheap enough for my students, the LLMs used here are relatively small and may not produce output of the same quality as larger models from Open AI and other global players. In addition, context lengths and token generation limits of Glif models seem to be smaller than those of "professional" LLMs.

Nevertheless, these components show what is possible and how it may be achieved - perfect for personal experiments and application prototypes. And if you need to enhance the blocks provided by the original glif-interfaces: you are free to fork your own variant and prepare your own Glifs with your own LLMs or LLMs of other providers using your own account at any time, otherwise keeping the original application the same.

Installation

glif-interfaces may be used as an ECMAScript module (ESM), a CommonJS or AMD module or from a global variable.

You may either install the package into your build environment using NPM with the command

npm install glif-interfaces

or load the plain script file directly

<script type="module" src="https://rozek.github.io/glif-interfaces/dist/glif-interfaces.esm.js"></script>

Access

How to access the package depends on the type of module you prefer

• ESM (or Svelte): import { GlifRunner,LanguageOfText } from 'glif-interfaces'
• CommonJS: const GlifInterfaces = require('glif-interfaces')
• AMD: require(['glif-interfaces'], (GlifInterfaces) => {...})

Alternatively, you may access the global variable GlifRunner directly.

Usage within Svelte

For Svelte, it is recommended to import the package in a module context. From then on, its exports may be used as usual:

<script context="module">
  import { GlifRunner,LanguageOfText,TranslationOfTextInto } from 'glif-interfaces'
</script>

<script>
  GlifRunner.APIToken = '...'
  ;(async () => {
    console.log(await LanguageOfText('Hello, World!'))
    console.log(await TranslationOfTextInto('Hello, World!','german'))
  })()
</script>

You may experiment with that code in the Svelte REPL

Examples

glif-interfaces contains numerous building blocks for AI agents, here are some examples (please note, that these examples may seem primitive - otherwise they would become too long - but do not underestimate what you may be able to do with these blocks in a real application with real-world inputs)

Language Detection and Translation

Here is how to detect the language of a given text and how to translate it into a different language (try yourself using the Svelte REPL)

import {
  GlifRunner,
  LanguageOfText,TranslationOfTextInto
} from "glif-interfaces"

;(async () => {
  GlifRunner.APIToken = '...'

  console.log(await LanguageOfText('Hello, World!'))
  console.log(await TranslationOfTextInto('Hello, World!','german'))
})()

The following languages are supported (these are the languages, the most common LLMs have been trained with. Your mileage may vary depending on LLM and language - "german" seems to work pretty well):

• arabic
• bengali
• bulgarian
• chinese
• danish
• german
• english
• estonian
• finnish
• french
• greek
• hebrew
• hindi
• italian
• korean
• croatian
• dutch
• portuguese
• spanish
• czech
• hungarian

If you want to experiment with that feature: here is the underlying Glif

Specification Review and Update

This example shows how to evaluate a given specification and how to update it by applying a list of changes. Please note, that while the original input is in german, the Glif "interface" (not the underlying Glif!) first translates it into english, then runs the Glif and finally translates the output back into german - because LLMs can handle english much better than any other language (try yourself using the Svelte REPL).

if you don't understand the german input: the first example (and its Glif) give you a possibility to translate it into your favourite language

import {
  GlifRunner,
  ReviewOfSpecification, SpecificationUpdatedUsing,
} from "glif-interfaces"

;(async () => {
  GlifRunner.APIToken = '...'

  console.log(await ReviewOfSpecification(
    'verlangt wird ein Pseudo-Number Generator in TypeScript, der in ' +
    'einer Browser-Umgebung funktioniert'
  ))
  console.log(await SpecificationUpdatedUsing(
    'verlangt wird ein Pseudo-Number Generator in TypeScript, der in ' +
    'einer Browser-Umgebung funktioniert',
    'Verwende einen "Linearen Kongruenzgenerator" (LCG). Vervollständige die ' +
    'Spezifikation so, daß eine einfache und sinnvoll einsetzbare Funktion ' +
    'entsteht. Wähle insbesondere vernünftige Konstanten für den LCG'
  ))
})()

You may repeatedly review and enhance your specification until either you or the AI seems confident enough that the specification may be implemented.

Code Generation

Given a specification, some additional requirements (such as a certain code style or the libraries to be used) and an already existing implementation (if any) can be used to generate (or update) the corresponding JavaScript or TypeScript code (try yourself using the Svelte REPL).

if you don't understand the german input: the first example (and its Glif) give you a possibility to translate it into your favourite language

import {
  GlifRunner,
  LanguageOfText, TranslationOfTextInto,
  JavaScriptImplementationOf,
} from "glif-interfaces"

;(async () => {
  GlifRunner.APIToken = '...'

  console.log(await JavaScriptImplementationOf(
`
Ein Pseudo-Zufallszahlengenerator in JavaScript wird benötigt, der in einer Browserumgebung funktioniert. Der Generator sollte einen "Linear Congruential Generator" (LCG) mit den folgenden Konstanten verwenden:

Modulus (m): 2^32
Multiplikator (a): 1664525
Inkrement (c): 1013904223
Startwert (x0): 12345

Die Funktion sollte einen optionalen Startwert-Parameter akzeptieren und eine Sequenz von Pseudozufallszahlen zurückgeben.
`, 'Vermeide Semikolons soweit möglich. Der fertige Code sollte ohne Dependencies auskommen und sofort einsetzbar sein', ''
  ))
})()

TypeScript code generation works in a similar way (see the related example in the Svelte REPL)

Code Review

Given a JavaScript or TypeScript code snippet, this example analyzes it and identifies actuel or potential problems (try yourself using the Svelte REPL).

import {
  GlifRunner,
  ReviewOfJavaScript,
} from "glif-interfaces"

;(async () => {
  GlifRunner.APIToken = '...'

  console.log(await ReviewOfJavaScript(
`
function lcg(x0 = 12345) {
  const m = 2 ** 32
  const a = 1664525
  const c = 1013904223

  return function() {
    x0 = (a * x0 + c) % m
    return x0
  }
}
`, 'ignoriere fehlende Kommentare', 'de'
  ))
})()

if you don't understand the german input: the first example (and its Glif) give you a possibility to translate it into your favourite language

The TypeScript code review works in a similar way (see the related example in the Svelte REPL)

(more to come)

API Reference

(t.b.w)

Build Instructions

You may easily build this package yourself.

Just install NPM according to the instructions for your platform and follow these steps:

1. either clone this repository using git or download a ZIP archive with its contents to your disk and unpack it there
2. open a shell and navigate to the root directory of this repository
3. run npm install in order to install the complete build environment
4. execute npm run build to create a new build

You may also look into the author's build-configuration-study for a general description of his build environment.

License

MIT License