Repository navigation
Releases
Releases · codeonym-oss/design-skills
v0.1.1
0.1.1 (2026-09-25)
Build & Image
- deps: Bump actions/checkout from 5 to 7 (#3) (c56c70f)
- deps: Bump actions/download-artifact from 4 to 8 (bdbfc0e)
- deps: Bump actions/upload-artifact from 4 to 7 (#4) (ae1eda7)
- deps: Bump docker/build-push-action from 6 to 7 (#1) (a433ab2)
- deps: Bump docker/login-action from 3 to 4 (#2) (78d81b7)
Images
docker pull ghcr.io/codeonym-oss/design-skills:0.1.1-core
docker pull ghcr.io/codeonym-oss/design-skills:0.1.1-full
v0.1.0
First public release, copied from the author's personal Arch workstation skills and made distro-agnostic and container-first.
Added
- Sandbox images
ghcr.io/codeonym-oss/design-skills:<version>-{core,full}on Ubuntu 26.04 LTS (amd64 + arm64):
core ≈ 1.2 GB (ImageMagick 7, Inkscape 1.4, ffmpeg 8, Ghostscript, Poppler, optimizers, exiftool, FontForge, fonts,
FOGRA/ISO Coated ICC profiles), full ≈ 2.9 GB (+ Blender 5, GIMP 3.2, G'MIC, Krita 6, Scribus 1.6, darktable 5, MLT, Mesa, Xvfb, Noto CJK). - Dispatcher (
lib/ds.sh): every skill script re-runs itself in the pinned image with the same paths; mounts
path arguments outside the working dir; runs as the calling user; VAAPI GPU passthrough; Docker or Podman;DS_RUNTIME=nativeescape hatch. dsCLI: run scripts,execany tool,shell,doctor,list,pull,image,test,version.design-sandboxskill documenting the environment.- Project fonts:
./fonts(andDS_FONTS) are visible to every tool in the container. - Distro-agnostic checks with install hints for apt, pacman, dnf and Homebrew (
lib/packages.tsv). print-layout/new-print-doc.sh --brand '#hex'; Scribus runs under Xvfb when headless.photo-editing/batch-develop.shaccepts a.dtstylefile path.- Test suite: 46 unit tests (bash 5 and 3.2) and 54 integration tests run inside the images; CI builds and tests
every variant × architecture before publishing.
Changed
- All skills and guides rewritten for any OS: no Arch/AUR commands, no machine-specific hardware, OBS guidance for
Windows/macOS/Linux, keystroke/zoom tools per platform. convert-model.shpasses paths to Blender as arguments (file names with quotes work).
Fixed (vs. the upstream Ubuntu packages)
- Blender's glTF Draco compression: Ubuntu ships the encoder under a CPython-tagged file name the add-on doesn't
load; the image links it so--dracoactually compresses. font-specimen.shno longer relies on ImageMagick@filereads, which common security policies forbid.
Images
docker pull ghcr.io/codeonym-oss/design-skills:0.1.0-core
docker pull ghcr.io/codeonym-oss/design-skills:0.1.0-full