Feat: LLM Chat and Nightlight shader #155

Open

zach wants to merge 21 commits from feat/llm-chat-tab into main

pull from: feat/llm-chat-tab

merge into: :main

:main

:feat/llm-chat-tab

:fix/lockscreen-font

:module/wifi

:fix/cli-typecheck

:feat/plugin-loading

:submenu-popouts

Conversation 4 Commits 21 Files Changed 136

+13940 -1167

zach commented 2026-08-31 18:57:51 +02:00

Owner

Copy Link

Copy Source

What & why

New chat interface in sidebar

The sidebar now has two tabs, notifications and chat. The chat is a frontend for whatever endpoint API the user sets in Config.llm.endpoint. Currently hardcoded-ish for llama.cpp endpoints but will generalize.

Tools

The chat has these tools defined:

• WebFetch (self-explanatory)

New nightlight shader

Hyprland plugin + C++ QML extension combo to support a custom shader implementation for nightlight which hooks Hyprland's render pipeline to insert itself after the buffer has been mirrored, meaning the shader doesn't shouldn't show up in screen captures.

New config options

• Config.display.nightlight.useNative
• Config.display.nightlight.fadeDuration

Fade duration is not available when useNative == false due to insufficient support from hyprsunset.

Backend config migration

We now have a reliable way to migrate objects in the config file, preserving the user's values cleanly. Can be found in Plugins/ZShell/Config/migration.{c,h}pp.

Usage

// migration.cpp
const QList<ConfigMigrationRule>& ConfigMigrations::rules() {
	static const QList<ConfigMigrationRule> s_rules = {
		migrate("old.object.path", "new.object.path"),
	};
	return s_rules;
}

Once migrated, rules are a no-op and can stay indefinitely (though they shouldn't.)

Related issue

No response

Type

New feature

Tested

• Build succeeds
• Existing tests pass
• Manually verified

Breaking changes?

No response

Checklist

• CI passes (format, lint, typecheck)
• Clean commits
• No new warnings
• Docs updated if needed

### What & why ## New chat interface in sidebar The sidebar now has two tabs, notifications and chat. The chat is a frontend for whatever endpoint API the user sets in `Config.llm.endpoint`. Currently hardcoded-ish for `llama.cpp` endpoints but will generalize. ### Tools The chat has these tools defined: - WebFetch (self-explanatory) ## New nightlight shader Hyprland plugin + C++ QML extension combo to support a custom shader implementation for nightlight which hooks Hyprland's render pipeline to insert itself after the buffer has been mirrored, meaning the shader ~~doesn't~~ shouldn't show up in screen captures. New config options - `Config.display.nightlight.useNative` - `Config.display.nightlight.fadeDuration` Fade duration is not available when `useNative == false` due to insufficient support from hyprsunset. ## Backend config migration We now have a reliable way to migrate objects in the config file, preserving the user's values cleanly. Can be found in `Plugins/ZShell/Config/migration.{c,h}pp`. ### Usage ```cpp // migration.cpp const QList<ConfigMigrationRule>& ConfigMigrations::rules() { static const QList<ConfigMigrationRule> s_rules = { migrate("old.object.path", "new.object.path"), }; return s_rules; } ``` Once migrated, rules are a no-op and can stay indefinitely (though they shouldn't.) ### Related issue _No response_ ### Type New feature ### Tested - [x] Build succeeds - [ ] Existing tests pass - [x] Manually verified ### Breaking changes? _No response_ ### Checklist - [ ] CI passes (format, lint, typecheck) - [x] Clean commits - [ ] No new warnings - [x] Docs updated if needed

zach added 15 commits 2026-08-31 18:57:52 +02:00

initial commit for llm chat tab in sidebar 47bab21e22

add removal of chat sessions, auto title + icon, proper stick-to-bottom as response is streamed 9657e5092a

start initial work on detaching chat to floating window 37c6a9986e

better chat list view + anchor to bottom. sqlite db for chats + LIFO-ordered aae3757daf

add markdown parsing + latex + tree-sitter highlighting for codeblocks in llm responses e18875a752

async parsing + tex caching 3946541f87

async parsing + tex caching 78988700be

rotated view + wheel inverter c33dbc147f

better scrollbar in codeblocks + enter anim for delegates 293bf67e09

expandable dialogs for color scheme, fonts and scheduling modes 8a66c3b064

better support for opening chat in floating window 327270f3fa

fab menu and model selector 703d804746

add nightlight hyprland plugin + rename hyprsunsetmanager -> nightlightmanager to control both d4e936a7e2

fix: shader fade animation linear curve 487615f482

feat: config object migration path and add new options to settings dbb51ceadd

zach added 1 commit 2026-08-31 19:02:30 +02:00

chore: format llm c files 52feb6006a

zach added 1 commit 2026-08-31 19:05:27 +02:00

chore: format llm c files c5a089adaf

zach added 1 commit 2026-08-31 19:06:56 +02:00

chore: format llm c files cef91caebc

AramJonghu commented 2026-08-31 19:24:37 +02:00

Collaborator

Copy Link

Copy Source

AI concern

My main concern would be if the chat window is an opt-in setting. I would personally think many users are not waiting for an llm integration into a quickshell project; I know I am not.

Will this be an opt-in feature? Meaning that the user would need to go into settings or the config file to enable the front-end to even show the chat window.

Build failure

It seems build is failing due to missing libcmark-gfm. I'm not sure what package this is as in artix repo + arch extra and multilib as I can't find it. I cannot find it in https://archlinux.org/packages or https://aur.archlinux.org either.

### AI concern My main concern would be if the chat window is an opt-in setting. I would personally think many users are not waiting for an llm integration into a quickshell project; I know I am not. Will this be an opt-in feature? Meaning that the user would need to go into settings or the config file to enable the front-end to even show the chat window. ### Build failure It seems build is failing due to missing `libcmark-gfm`. I'm not sure what package this is as in artix repo + arch extra and multilib as I can't find it. I cannot find it in https://archlinux.org/packages or https://aur.archlinux.org either.

zach commented 2026-08-31 21:43:58 +02:00

Author

Owner

Copy Link

Copy Source

There will be a config option to disable it, like everything else. It'll most likely be enabled by default though so that users can see that it exists without going deep into settings and stumble upon it.

I would personally think many users are not waiting for an llm integration into a quickshell project; I know I am not.

I disagree with that statement wholeheartedly, I don't think that LLM integration and Quickshell projects are mutually exclusive, in fact there's already multiple Quickshell configs that implement an LLM chat interface.

The package is just called cmark.

There will be a config option to disable it, like everything else. It'll most likely be enabled by default though so that users can see that it exists without going deep into settings and stumble upon it. >I would personally think many users are not waiting for an llm integration into a quickshell project; I know I am not. I disagree with that statement wholeheartedly, I don't think that LLM integration and Quickshell projects are mutually exclusive, in fact there's already multiple Quickshell configs that implement an LLM chat interface. The package is just called `cmark`.

AramJonghu commented 2026-08-31 22:00:40 +02:00

Collaborator

Copy Link

Copy Source

I meant to say opt-out, not opt-in. I'd prefer opt-in personally.

I mean to say that a general user using a desktop environment or a desktop like shell would unlikely want llm chat integration. There are no statistics I have found of this, but reading people leaving windows generally do not like AI. Granted; Microsoft shoves Copilot everywhere in Windows.

It's up to you. I'm okay with it so long it can be hidden completely (apart from the setting to be able to toggle it to enable the feature). Since it is a config option, all good for me.

I meant to say opt-out, not opt-in. I'd prefer opt-in personally. I mean to say that a general user using a desktop environment or a desktop like shell would unlikely want llm chat integration. There are no statistics I have found of this, but reading people leaving windows generally do not like AI. Granted; Microsoft shoves Copilot everywhere in Windows. It's up to you. I'm okay with it so long it can be hidden completely (apart from the setting to be able to toggle it to enable the feature). Since it is a config option, all good for me.

AramJonghu added 2 commits 2026-08-31 22:48:06 +02:00

chore: format python codeicons 51dc0fc026

fix(Docker): add required deps to dockerfile c9f7856982

AramJonghu force-pushed feat/llm-chat-tab from 24593472e3 to c9f7856982 2026-08-31 22:48:06 +02:00 Compare

zach added 1 commit 2026-08-31 22:54:55 +02:00

fix: font family 2150859847

AramJonghu commented 2026-08-31 23:28:43 +02:00

Collaborator

Copy Link

Copy Source

c9f7856982 fixes the missing dependencies in the Dockerfile. Issue is that the build still fails, maybe related to ci/CMakeLists mismatch with new dependencies.

[c9f7856982](https://git.zach-dev.cc/zach/z-bar-qt/commit/c9f7856982d970955ac87db29098ac6eb929f81d) fixes the missing dependencies in the Dockerfile. Issue is that the build still fails, maybe related to ci/CMakeLists mismatch with new dependencies.

Some required checks failed

Hide all checks

JS/TS / fmt (pull_request) Successful in 19s

Details

JS/TS / lint (pull_request) Successful in 18s

Details

Rust / fmt (pull_request) Canceled after 0s

Details

Rust / clippy (pull_request) Canceled after 0s

Details

Rust / build (pull_request) Canceled after 5m56s

Required

Details

Python / static (pull_request) Canceled after 6m44s

Details

Python / verify (pull_request) Canceled after 6m40s

Required

Details

C++ / fmt (pull_request) Successful in 8s

Details

C++ / clang-tidy (pull_request) Canceled after 7m8s

Details

C++ / build (pull_request) Failing after 1m24s

Required

Details

Some required checks were not successful.

This pull request doesn't have enough required approvals yet. 0 of 1 official approvals granted.

You are not authorized to merge this pull request.

This branch is out-of-date with the base branch

View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.

git fetch -u origin feat/llm-chat-tab:feat/llm-chat-tab

git checkout feat/llm-chat-tab

Sign in to join this conversation.

Reviewers

No Reviewers

Labels

Clear labels

bug

Something isn't working

documentation

Improvements or additions to documentation

duplicate

This issue or pull request already exists

enhancement

New feature or request

good first issue

Good for newcomers

help wanted

Extra attention is needed

invalid

This doesn't seem right

question

Further information is requested

wontfix

This will not be worked on

No labels

Milestone

No items

No Milestone

Projects

Clear projects

No projects

Assignees

Clear assignees

AramJonghu Inorishio zach

No Assignees

2 Participants

Notifications

Subscribe

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: zach/z-bar-qt#155