threadcat; warc2zip; quilt.id
We have a diverse and eclectic mix of resources today, covering everything from slurping up content from long Fediverse blathers to making it easier for regular folks to work with Common Crawl data, to yet-another fancy identicon implementation.
threadcat

As most of you know all too well, I hang out on Mastodon quite a bit, and both post and consume my fair share of threads. I primarily use Mona when posting threads but usually end up doing something pretty hacky to save content from threads into my local Joplin setup. That is, until I came across threadcat (CB).
This utility takes the URL to a Mastodon post, and converts the thread (all children by the same author) to Markdown in with a YAML header (which is standard format for static-site blog posts). It also downloads all included media files, and their alt texts. NOTE: given the name of the tool, I feel compelled to ensure folks know that a thread can be a single post
You will need a Rust toolchain handy since the Codeberg repo has no pre-built binaries and a quick brew info shows no recipe for it for macOS (I just assume Linux/BSD folks have Rust lying around these days).
Usage is simple (threadcat URL), but you should consider using --output if you want something besides index.md as the destination file and I highly recommend starting work in a subdir just in case there are attachments.
warc2zip

The WARC (Web ARChive) file format bundles multiple web resources into a single archive file. It can be replayed with tools like ReplayWeb.page or served by archive sites like the Wayback Machine. It’s an extension of an older ARC format to support harvesting, access, and exchange needs of archiving organizations — including metadata, duplicate detection, and later transformations. Its HTTP/1.0-inspired structure (headers, CRLF delimiters) makes it easy for crawlers to implement.
First specified in 2008, WARC is the standard for web archiving in most national libraries. WACZ, released in 2021, wraps WARC files in a ZIP container for random access and indexing. It keeps the WARC as it is and just adds said wrapper.
warc2zip (GH) is a new format + utility from the Common Crawl team. Each response becomes a file named with a counter and an extension intuited from Content-Type, and everything else goes into CSVs, including manifest.csv, manifest.jsonl, the request and response header CSVs, metadata.csv, and warcinfo.csv.
The blog post didn’t say how the ZIP is compressed, so I yanked and read the source. It opens the output with zipfile.ZIP_DEFLATED and writes payloads from warcio’s content_stream(). That means the bodies have chunked encoding and content-encoding already removed, and the original WARC record bytes never go into the ZIP. A warc2zip output can’t be replayed, isn’t a WACZ, and you can’t rebuild the source WARC from it byte for byte.
Common Crawl’s stated purpose for warc2zip is accessibility. Their post says the utility turns WARC archives into ZIP files that “can be inspected with ordinary tools,” such as spreadsheets, so users don’t need to know the WARC format or wield a Python library. The authors frame the goal as making the corpus accessible “at every level of technical skill.”
So, the intended users are non-specialists. The post’s examples are “a historian” wanting PDFs from government sites or “a linguist” wanting Swahili-tagged pages, while keeping crawler metadata intact. The metadata is pretty important to help preserve the full provenance, including the source WARC URI, offset and length, so any capture can still be traced back to its original record.
It’s super cool to see the WARC ecosystem is alive and well in 2026, and we’ll cover some more modern WARC utilities and practices in upcoming Drops.
quilt.id

An identicon is a small geometric image generated deterministically from a piece of data — usually a hash of an identifier like an IP address, username, or public key. So, the same input always produces the same pattern, and acts as a sort of “visual fingerprint”: you can recognize an entity at a glance and spot when two things are actually the same.
The original concept comes from Don Park’s 2007 “Identicon” proposal, which rendered a 5×5 symmetric grid from an MD5 hash. Since then the idea has spread into many forms including GitHub-style avatars (the default profile pictures for accounts with no avatar set), network/security tooling (many dashboards render identicons for IPs or hosts so analysts can visually distinguish them in dense logs), and even email clients (some show identicons next to unknown senders as a trust signal).
Key properties include determinism (no stored image needed), collision-resistance (enough for casual visual distinction), and inability to reverse back to the source data. Please remember that these identicons are a convenience, not a security control — two different inputs can look similar to a human eye, so never use visual matching alone to confirm identity.
A new identicon format recently came into the field — quilt.id. It runs your text through SHA-1, a math function that mashes any input into a long, scrambled-looking number. Two inputs that differ by one letter come out totally different. Only the first 48 of the 160 bits get used. The first part of those bits picks a set of three colors. The next 30 bits fill 15 cells, two bits per cell: 00 means empty, and 01 through 11 pick one of the colors. The left two columns are then mirrored onto the right, which makes the picture symmetric. That gives a 5×5 grid where you only “paint” the left half plus the middle column, and the right half is a mirror image. The mirroring is why the quilts look like faces or little creatures and not random noise, since our little grey cells love symmetry. The corner cells are always left empty so the quilt fits inside a circle crop.
I didn’t see a repo link on the site, and am loathe to hand over an HTTP request to a third-party if I don’t have to, so I implemented the algorithm (it’s trivial) in Go which is at https://rud.is/git/quilt/. It works as a standalone CLI and also sports a REST API (via the serve command), which you can reverse proxy behind Caddy (or other, inferior setups).
FIN
Remember, you can follow and interact with the full text of The Daily Drop’s free posts on:
- Mastodon via
@dailydrop.hrbrmstr.dev@dailydrop.hrbrmstr.dev - Bluesky via
<https://bsky.app/profile/dailydrop.hrbrmstr.dev.web.brid.gy>
☮️
Leave a Reply