Usage

The web interface

1. Open app.argonfetch.dev - or your own instance, at http://localhost:8080 unless you moved it.
2. Paste a media URL and press Enter.
3. ArgonFetch resolves the link and shows the title, author and cover, with the qualities the source actually offers.
4. Pick one. The file downloads through your browser like any other.

Nothing is queued and nothing is kept - the stream is proxied through as it is fetched.

From the command line

The API is a handful of GET endpoints, so curl is enough. Resolve first, then stream the rendition you want.

The examples below run against the hosted instance, so they work as they stand. Swap the host for your own - http://localhost:8080 by default - if you are self-hosting.

Resolve a link:

curl -s "https://app.argonfetch.dev/api/Fetch/GetResource?url=https://www.youtube.com/watch?v=dQw4w9WgXcQ"

The response carries a key for each available rendition (see the API reference for the full shape).

Download that rendition:

curl -L "https://app.argonfetch.dev/api/Stream/Media/<key>" -o track.webm

Ask for MP3 instead of the source container:

curl -L "https://app.argonfetch.dev/api/Stream/Media/<key>?format=mp3" -o track.mp3

When video and audio are separate streams, the rendition's urlType is Combined and the muxing endpoint is the one to call:

curl -L "https://app.argonfetch.dev/api/Stream/Combined/<key>" -o video.mp4

Formats

Audio is delivered in whatever container the source uses - Opus in WebM for YouTube - because passing the original bytes through is faster and lossless. Every rendition states its mimeType and fileExtension, so you always know what you are getting.

The one thing pass-through cannot do is carry tags, which makes this a real choice rather than a formality:

You want	Do this	You get
The audio exactly as the source has it	Take the audio rendition as-is	Byte-identical, no re-encode, no tags inside the file
A tagged file your music player will file correctly	Add ?format=mp3 to the stream URL	Re-encoded to MP3 with the title and artist written in
Video	Take a video rendition	MP4, tagged while it is muxed

Sources in a container ArgonFetch does not recognise are converted to MP3 automatically, so those arrive tagged either way.

Filenames and tags

Every download names itself. Both stream endpoints send a Content-Disposition built from the media, so the file lands as Artist - Title.ext rather than a cache key - in a browser, in curl -OJ, and in anything else that reads the header:

Content-Disposition: attachment; filename="Rammstein - Sonne.webm"; filename*=UTF-8''Rammstein%20-%20Sonne.webm

The name is written twice on purpose. The quoted form stays ASCII so clients that read only that still get something sensible, and filename* carries the real name for everything else - which is what makes a title in Japanese or Cyrillic survive the trip.

Tags inside the file are a separate matter, and only anything ArgonFetch converts gets them:

• Converted output - MP3, and the muxed MP4 - carries the title and artist internally, written while the file is being built.
• Pass-through carries whatever the source shipped with, which is usually nothing. Tagging it would mean remuxing, and that would cost the exact byte length the response already promised and the range requests a client may be making against it. The filename carries the information instead.

There is no cover art in either case. Embedding a picture needs a seekable output, and these responses are pipes.

Trying the API without curl

Every endpoint has a page under API with a playground that sends the request for you, against the hosted instance or your own. The reference UI is also served at /scalar, though only when the instance runs with ASPNETCORE_ENVIRONMENT=Development - so on a local build, not on app.argonfetch.dev. The schema is checked in at openapi.json either way, for generating clients without running anything.

A note on what you download

ArgonFetch does not check what you are allowed to keep. Make sure you have the right to download the content you point it at.

Next

• Supported platforms - what works today and what does not.
• API reference - every endpoint and response shape.