mark is a serverless shortlinking service powered by CloudFlare Workers, mostly for my own private use. The name comes from vim's marks, which let you jump to a position in a file (and is also conveniently short for bookmark). My motivation for making this was to have a short way of referring to links on paper without having to write the full URL out by hand. I also wanted a service that I was in control of, so services like bit.ly were out of the question (besides, this was pretty fun to make).
Sending a GET request to any path under https://mark.imranh.org
will attempt to find a link with that path.
Try it with the key EGFa
!
This is a "private" service, i.e. the intent here is that only one person is generating new links
(but obviously anyone can visit existing ones).
To create new shortlinks, send a POST request to /new
with a JSON object containing keys:
url
: The URL to be shortenedauthkey
: A password
This will output a new four-character base62 path that can be easily written down. This only allows for around 15 million links, but that should be plenty for one person :)
Note that there is no frontend or GUI, this is meant to be as simple as possible.
It's pretty easy to send POST requests from the command line, and this can be easily scripted.
Below is a script I made for myself to make managing links easier:
The backend is run by CloudFlare Workers, and links are stored in
Workers KV, a global eventually-consistent key-value store.
GET /:key
will redirect (HTTP 301) with the value of key
from the KV namespace.
POST /new
(with the correct authkey) will write a new random base62 key to that namespace
with the given link as the value.
This repo uses CloudFlare's wrangler to manage development and deployment, as well as pnpm to manage Node packages. It also uses itty-router to make this easier to write.
Since the wrangler.toml file will contain account-specific info (like account & zone IDs), I use 1Password's CLI to inject these into a template file.
I'd like to add an archive function to this eventually, so webpages with different types of media can be stored and retrieved, even if the original page becomes unavailable.