Hi Jaeme,

Thanks for submitting this package!  I've attached my revision 
below.

We don't sign off on our own patches in Guix.  Our Signed-off-by 
does not have the same DCO meaning as it does for, e.g., Linux. 
Ours is merely a stamp of approval by a committer.

+(define-public ani-cli
+  (package
+    (name "ani-cli")
+    (version "4.6")
+    (source
+     (origin
+       (method git-fetch)
+       (uri (git-reference
+             (url "https://github.com/pystardust/ani-cli")
+             (commit (string-append "v" version))))
+       (file-name (git-file-name name version))
+       (sha256
+        (base32 
"1ni9pzjb5qh87iz7c8252bx79qadr1qx6jnkqvvjcqrchh7q473a"))))
+    (build-system copy-build-system)

So…  uhm.

Apparently the copy-build-system doesn't acknowledge the existence 
of cross compilation, like, at all?

This is a bug in Guix, not you, but here I switched to the 
gnu-build-system to avoid it.

+                       (wrap-program (string-append #$output 
"/bin/ani-cli")
+                         `("PATH" ":" prefix
+                           ,(map bin
+                                 (list "aria2c"
+                                       "curl"
+                                       "ffmpeg"
+                                       "fzf"
+                                       "grep"
+                                       "mpv"
+                                       "sed"
+                                       "uname"
+                                       "yt-dlp")))))))))

I added "tput" to this list.  While the script appears to be 
usable without it, it spits out ugly errors and might misbehave.

+    (home-page "https://github.com/pystardust/ani-cli")
+    (synopsis "CLI to browse and watch anime")
+    (description
+     "ani-cli is a command-line interface (CLI) to browse and 
watch anime.  It
+provides an easy way to stream shows using allanime as a source. 
There are
+different features such as episode browsing, history tracking, 
streaming at
+multiple resolutions and much more depending on what programs the 
user has
+installed.")

I took the liberty of rewriting this a bit, including my previous 
@uref suggestion.

+    (inputs (list aria2
+                  coreutils
+                  curl
+                  ffmpeg
+                  fzf
+                  mpv
+                  yt-dlp))

Besides ncurses for tput, I added grep and sed for the wrapper. 
This will matter when the package cross-builds: you don't want the 
script calling the build-time ‘native’ grep at run time.

I also added bash-minimal at ‘guix lint’'s suggestion.

We conventionally put fields in this order:

  source
  build-system
  arguments
  native-inputs
  inputs
  propagated-inputs
  home-page
  synopsis
  description
  license

This is not a hard rule and you'll find variants, but I moved the 
inputs to their conventional location above the Metadata Zone.

+    (license license:gpl3)))

As noted in #guix, I changed this to GPL3+ due to the ‘or (at your 
option) any later version’ wording.

Here's a random Deep Guix Thing that I'm too tired to explain 
further: because this script invokes curl, and to avoid 
propagating curl, I added a copy of one of our curl package's 
native-search-paths.  This makes the script work inside a --pure 
guix shell, as long as you include nss-certs.

I do wonder: just how legal is this script, and the one Web site 
to which it's tightly bound?

Kind regards,

T G-R