I want you to explore https://blog.jetbrains.com/kotlin/2026/01/the-journey-to-compose-hot-reload-1-0-0/ and anything that you find any concepts that you find you can further explore in the web and help me plan out because right now the way we have hot load implemented the hot load does work but it's not like instant it takes a second or two you can read our code our current code to understand why is that happening and then propose solution of whatever these guys are using these jet brains guys are using if we if it's if it's possible to use it then we probably want to use it in our case as well the same mechanism so we can have a faster hot reload like an instant hot reload any adjustments any breaking changes that we have to do I am okay with that if you have any questions please ask.
implement 01xcaa3f-d7ce-7841-8fe6-83ef644152b5.md . ask questions if any. check web if you want and temp scripts if you need to confirm something. project-to-refer is in path/to/project-to-refer if you need . i also want to you create todos in todo.md with phases and check out items as you progress
to self: 01xcaa3f-d7ce-7841-8fe6-83ef644152b5.md is the whole convo exchange not just plan file in codex
that works. thanks. i want you look at last commit and think and tell why all those changes are necessary and which ones of them are redundant, fallback, not useful or not in use anymore so we can drop them. please think really hard and check web if you need
can you check the code and think really hard and check web if you need to tell why it could be happening? you can run local tests if you want to confirm hypothesis or to grab findings
better ui llm:
❯ Great, thanks. Now I want you to create a fresh plan to show this in the UI. use tailwind css . Also add in the fact that this JSON file may not be there because the command would not have been on. So assume the fact that this could have happened and also incorporate it. You know, that I think suppose the UI should tell you that run this please, run this command please and then refresh something like that. also Don't add any flare in this site. This should be as simple as possible. I do want a search though, so I'm not sure, use minisearch or something, the main things I want to search for is by searching searching through like filtering by by an environment, like if the deployments have happened essentially NOT_EXECUTED is not there for environment for that environment yeah so yeah so show only those then also a way to essentially if you i'm not totally sure how you're going to demonstrate show all of this but if what a way to essentially a way to toggle all those blocks for all the builds in which NO_EXECUTED is there so I can easily see where the build has flown through. I would also prefer a way to mention a tag SCM_TAG and see all the places where it's deployed. see @data/history.json to understand the structure first
Also search the web to understand like what sort of policies that LLMs make when they create UI like for example showing a purple gradient and all because we want to avoid that and then also explore like good designs I'm not sure maybe on dribble or in spotted in prod or check some design studios like not design studios because they show a lot of there using 3js and all we don't want that we want a simple great clean website for this but yeah take some inspirations from the web. I'm not sure there are other places as well from which you can take the inspiration so take the inspiration from there.
- prepare an implementation-ready patch spec with exact edit points so it can be applied immediately once execution mode is enabled.
- cite relevant references (paths, links)
- ask questions if any
https://github.com/JetBrains/kotlin-desktop-toolkit So we have this Kotlin desktop toolkit and I wanted to create a sample program in this folder using it. In case if you want to consult the code to understand the project better, you can use these resources:
- https://github.com/JetBrains/kotlin-desktop-toolkit
- https://deepwiki.com/JetBrains/kotlin-desktop-toolkit
- this local folder which has the code: /path/to/kotlin-desktop-toolkit-main
prepare an implementation-ready patch spec with exact edit points so it can be applied immediately once execution mode is enabled.
cite relevant references in the plan and in the code comments if needed
no fallback or maintaining backwards compatibility pls, let's make proper breaking change
unrelated to chatgpt:
- AI sycophancy
- https://pi.dev/ and Worktrunk, also see this
- I want you to implement this and before you start implementing, if you have any questions, please ask.
- check web for inspo.
- remind it about creating python program for inspection And make it aware of jq tool for example
- qmd, feeding docs
- linter external program linter after modifying file for repetitive mistakes
- cursor has tools or something for the linter
- repetitive changes create programs and run them instead of directly modifying code
- making and run cmd itself and check and fix/correct (loop)
- using skills, hooks for custom linters?
- cursor cli?
- frontend design
/plugin marketplace add anthropics/claude-code/plugin install frontend-design@claude-code-plugins- https://x.com/trq212/status/2017024445244924382
- https://x.com/sanyuan0704/status/2019258745705099670
- use post hook or ghostty with cursor cli to receive audible notification
/initusing claude sonnet in cursor? check how composer's output look like- asking it to spawn agents for exploration
- pros: parallel search means less context consumption by main agent (i think)
- cons: feels to me tokens get consumed faster which is weird because i thought it would be mcp like. also i felt that final output by main agent is better when it is the only doing all the searches and all and not fanning/spawing out the agents to do the search
- dunno how true it is AGENTS.md outperforms skills in our agent evals
- if tired of copy pasting error output: https://github.com/mitsuhiko/vite-console-forward-plugin, rust-skia-live screenshot capability (or macos window screenshot capability in general + macos or apps own accessibility tree or dom/xml tree), terminal stuff like:
- tmux read
- that python lib armin ronacher mentioned that can execute commands event on interactive tui app i guess - pexpect in python - see youtube Armin Ronacher videos: Ubertool MCPs and lldb videos
- https://github.com/badlogic/terminalcp (via https://mariozechner.at/posts/2025-08-15-mcp-vs-cli/)
- https://aliou.me/pi/extensions/pi-processes/
- https://github.com/marcfargas/holdpty/
- "Think really hard about this" still does work (in cursor at least, not sure about "think hardest")
- Agents understand annotation on images You can send image of how ui looks like and another image with annotations (and maybe numbering for sequencing + later referencing). Also see:
- tldraw
- https://benji.org/annotating
- https://x.com/benjitaylor/status/2024622252424384995?s=20
- https://x.com/benjitaylor/status/2019454660575146062?s=20
- https://www.youtube.com/watch?v=r2e4J8g616Q
- https://x.com/decodetool and https://x.com/seflless
- https://www.youtube.com/watch?v=riFFyTRljic and https://www.youtube.com/watch?v=s7BZz03icnY
local and remote isolated vms:
- local
- lxd https://github.com/canonical/lxd https://github.com/canonical/lxd-ui
- broader than lxd/ubuntu: https://linuxcontainers.org/incus/
- canonical multipass
- seven up
- remote
- web Search
- https://git.lolcat.ca/lolcat/4get, https://4get.ca/instances
- https://github.com/benbusby/whoogle-search + its inbuilt json results mechanism
- gets rate limted really quickly
- elinks + Brave search
- https://www.mojeek.com/ is good alternative to Brave but it doesn't crawl Reddit which is a dealbreaker
- Unfortunately https://www.startpage.com/ gives captcha quite easily
- web fetch
- for certain sites:
- https://farside.link/
- for HN https://api.hackerwebapp.com
- for reddit with
.json?raw_json=1and other query params (public API is still heavily rate limited though) - https://nitter.tiekoetter.com (this one is without anubis captcha)
- https://www.npmjs.com/package/@edmundmiller/pi-scurl
- elinks
- https://github.com/miso-belica/jusText
- https://github.com/adbar/trafilatura or its Go port
- https://github.com/mozilla/readability or its Go port
- https://markdown.new/ and https://defuddle.md/
- None of these will actually give result of SPA page. if you could able to directly tool call Codex's web fetch then you'd get the right output or use https://developers.cloudflare.com/browser-rendering/
- for certain sites:
- web fetch + web search
in cursor:
Shared Knowledge
View team-wide Rules, Commands, and Hooks.