Patch any Omarchy shell plugin and keep your changes in a git repo. Hit Patch in the PlugPatcher panel, edit the code, and your version loads while the original stays installed and keeps updating — pull upstream updates in with your edits on top, and open a pull request when you're ready.
omarchy plugin add https://github.com/mxKeaton/plugpatcher --enableNothing else — the plugin ships the CLI it runs.
Put your patched plugins back, then remove the widget and its data:
CLI=~/.config/omarchy/plugins/io.github.mxkeaton.plugpatcher/bin/plugpatcher
"$CLI" list # what is being edited
"$CLI" revert <plugin-id> # repeat for each
omarchy plugin remove io.github.mxkeaton.plugpatcher --yes
rm -rf ~/.local/share/plugpatcher ~/.local/state/plugpatcher ~/.config/plugpatcherEverything lives in your home directory; /usr/share/omarchy is never touched.
Each plugin in the panel is a card:
- Patch — start editing a plugin.
- AI — open your coding agent in the code.
- Editor / Browse — open the code in your editor / file manager.
- Update — pull upstream updates in and re-apply your edits on top.
- Source — open the plugin's GitHub page.
- Send PR — Manual PR (GitHub's form) or AI PR (writes the text); Cancel PR withdraws it.
- Delete — remove the original, the patched copy, or both.
- switch — load patched ⇄ original (bar widgets restart the shell, ~3–4 s).
Update & Sync runs Omarchy's plugin update and then rebases your patched
plugins; Sync does just the latter. plugpatcher adopt <id> brings an
existing plugin repo (with your commits) under PlugPatcher.
Omarchy (Quickshell) · git, jq, python3 · gh (only for PRs) · systemd
(systemd-run)
Copy skills/plugpatcher/SKILL.md into your agent's skills directory (e.g.
~/.pi/agent/skills/plugpatcher/) to teach it the workflow.
MIT
