Thanks for the feedback, will tweak the prompts.
108 karma · joined May 11, 2020
This one is pretty simple and works in linux/mac and similar terminals. Instead of writing the shell command, you provide english instructions of what you want to achieve, ChatGPT translates it to shell syntax, and executes it after you confirm by hitting enter.
You also need to define OPENAI_API_KEY env variable.
Will add instructions in the github README.
If I understood correctly, you want a more advanced system, which would go search the internets for a given topic, collect a set of webpages around it, and come up with an organised presentation of information, in form of an article or a book.
It could also possibly suggest a set of topics given the users interests.
Sounds like an interesting and much more advanced project.
gistxy.com allows you to do just that (GIst STatic proXY).
If you have a Gist URL like:
https://gist.github.com/tzador/80c83ed5f5a0bdf8c878c5478d7d8...
just replace the .github. part with .gistxy. to get GistXY URL:
https://gist.gistxy.com/tzador/80c83ed5f5a0bdf8c878c5478d7d8...
and you will get the contents of the Gist served with correct mime-types and a working micro-website.
It also renders the markdown files to HTML, so you can focus on reading or sharing just one markdown file.
For security reasons, each Gist is served on dedicated subdomain.
Not sure how useful this is, but it was quite fun project.
Written in https://kit.svelte.dev and deployed on https://vercel.com .
Due to rate-limits, it scrapes the gist data from the Gist website, instead of using the GitHub API.
Yes, the search button is not needed, since we live show the results which are searched in memory. Removed.
Wanted to keep the speedy results as you type, so instead of debouncing for the preview now the native emojis rendered as chars are shown, not images.
If there is much traffic, there must be ways to optimise image serving, unless you have other concerns.
Let me know please if you see other ways to improve this small project.