iiinova — where your digital self keeps evolving

Kevin Kern @kevinkern.dev · 29/01/2026

Dropbox → Nextcloud Github Runner → Local Web Hosting → Coolify + Hetzner Postman → bruno What I'm still keeping: Notion Vercel

010

Kevin Kern @kevinkern.dev · 29/01/2026

Over the last few weeks I've been moving away from a few cloud providers. I've wanted to do this for a while, not because they're bad, but because a lot of them are built for bigger teams/enterprise and I don't need most of those features. Slack → RocketChat Pandadoc → Docuseal

120

Kevin Kern @kevinkern.dev · 14/01/2026

6. told claude to migrate all the gemini 3 frontend code to vite, use package.json (aistudio code does inline imports) and port styles to tailwind/shadcn. + remove all hardcoded color schemas. 7. some more iterations with opus + gpt-5.2 (wire real data instead of mock data + add jira api) 8. done

010

Kevin Kern @kevinkern.dev · 14/01/2026

4. create a new tauri desktop app template 5. paste all the code into the boilerplate 6. add deepwiki & context 7 mcp

100

Kevin Kern @kevinkern.dev · 14/01/2026

Jira is so slow and bloated that it was worth to spend some time to build my own desktop app (that syncs incoming issues) How? 1. wrote a spec in gpt-5.2 2. paste spec to aistudio (Gemini 3 Pro) for frontend 3. paste spec to gpt pro for initial mvp backend code

110

Kevin Kern @kevinkern.dev · 05/01/2026

I mean, I really like Opus. It's like those friends who tell the best dressed up stories. And in your inner self, you know it's mostly just for the show.

000

Kevin Kern @kevinkern.dev · 05/01/2026

So peer review is back.

100

Kevin Kern @kevinkern.dev · 05/01/2026

That's another reason why GPT-5.2 reviews Opus-4.5. great at rushing through the codebase and identifying mistakes. At the same time, it's sloppy and sometimes stops halfway instead of following through. still overpromises, but with gpt-5.2 oversight we consistently land on high quality results.

100

Kevin Kern @kevinkern.dev · 03/01/2026

I've also added a skill that checks other agents' work and progress if we face a blocker.

000

Kevin Kern @kevinkern.dev · 03/01/2026

Since Codex is taking forever for larger codebases, I split my tasks across to 5-8 agents. the important part is that it's split by domain so you don't overwrite things. But Codex is pretty good at recognizing others' work and stops or asks if it should proceed.

110

Kevin Kern @kevinkern.dev · 21/11/2025

One tip: Cursors composer-1 model is pretty fast. Which means you could also try to run the above prompt via cursor CLI.

010

Kevin Kern @kevinkern.dev · 21/11/2025

was playing around with Codefetch and oracle to gather context and send hard tasks to GPT-5.1 Pro. here's a slash command I ended up with.

110

Kevin Kern @kevinkern.dev · 20/11/2025

little annoying that Codex removes things "for good". I'm telling it this from now on.

010

Kevin Kern @kevinkern.dev · 20/11/2025

Instead of switching between MCP servers like chrome devtools, context7, linear... you create a single workflow (e.g., frontend_tools) that orchestrates them. Your MCP client (Cursor, Claude Code) sees minimal needed tools.

100

Kevin Kern @kevinkern.dev · 20/11/2025

Just released Oplink.⚡️It combines multiple MCP servers into unified workflows.

100

Kevin Kern @kevinkern.dev · 20/11/2025

Nice addon for codex. Oracle sends your hard tasks to gpt 5 pro. Tip: You can pair it with codefetch to set up the context first.

000

Kevin Kern @kevinkern.dev · 12/11/2025

TIL: ChatGPT can create simple sound files for you (I was looking for free sounds for codex-1up).

010

Kevin Kern @kevinkern.dev · 11/11/2025

A fresh codex-1up update is out. - Quick install "npm install -g codex-1up" - Add custom sounds when a task is finished - Guided setup (easy for beginners) - Lot of cleanup, improvements - New updated profiles to start quickly

100

Kevin Kern @kevinkern.dev · 10/11/2025

in the next version of codex-1up, a new fast guided setup.

000

Kevin Kern @kevinkern.dev · 09/11/2025

A new stealth model, polaris-alpha, has appeared on openrouter. Time to give it a try.

000

Kevin Kern @kevinkern.dev · 07/11/2025

Multiple way to serve context. But you always should keep an eye on your context window and don't pollute it with unnecessary or wrong information. Thats your "job" as agentic engineer.

000

Kevin Kern @kevinkern.dev · 07/11/2025

- The agent is able to read linters - The agent has access to externals tools served via CLI/MCP - The agent has memory of your past decisions - The agent has access to tasks written in .md

100

Kevin Kern @kevinkern.dev · 07/11/2025

What does high context mean? - The agent is able to run tests against your codebase. - The agent has access logs - The agent has rules available (AGENTS .md) - The agents has access to semantic search (thats why its still useful to index your codebase) - The agents has relevant docs available

100

Kevin Kern @kevinkern.dev · 07/11/2025

To explain: With better models and a codebase that serves as context you can just talk to the model. The less context you have, the more specific your prompt should be.

100

Kevin Kern @kevinkern.dev · 07/11/2025

My earlier take is deprecated.

100

Kevin Kern @kevinkern.dev · 06/11/2025

I've updated today the tanstack starter project + agent rules. You can build now quickly web apps with one the best web dev stack available.

000

Kevin Kern @kevinkern.dev · 06/11/2025

You'll receive findings (high, medium, low), including an analysis of missing tests for the given plan.

000

Kevin Kern @kevinkern.dev · 06/11/2025

f you use composer-1 in cursor for coding, it's helpful to run a quick /code-review to identify any code smells. In my case I use a slash command with gpt-5. Usually you find some flaws in the final implementation.

100

Kevin Kern @kevinkern.dev · 05/11/2025

Finally I found a way to 10x codex! Sound on! 🔊

000

Kevin Kern @kevinkern.dev · 03/11/2025

The changelog contains a complete list of all updates.

000

Kevin Kern @kevinkern.dev · 03/11/2025

composer-alpha-1 is now my go model for fast ask/answer/edit tasks. GPT-5 is still excellent for solving complex, high-level problems (especially GPT-Pro and Deep Research), but it’s not ideal for quick edits. Using both models together has given me my most productive workflow so far.

110

Kevin Kern @kevinkern.dev · 03/11/2025

Summary: I've tested the new version for a few days and really enjoyed using the new composer model. Compared to Claude Sonnet and GPT-5, it feels closer to Sonnet but with much better code understanding and results inside Cursor. (Not talking about CC Sonnet)

100

Kevin Kern @kevinkern.dev · 03/11/2025

5/ Better Code Review You can now view all changes from Agent across multiple files in one place, without switching between files.

100

Kevin Kern @kevinkern.dev · 03/11/2025

The worktree option lets you build features in isolated branches. With Best-of-N, multiple agents run in parallel, and each git worktree has its own clone of the codebase to prevent conflicting changes.

100

Kevin Kern @kevinkern.dev · 03/11/2025

4/ Git worktree and best-of-n selection We have seen this in codex before. Cursor generates multiple versions of a task so you can pick the best one.

100

Kevin Kern @kevinkern.dev · 03/11/2025

3/ Browser Mode This is a good one for designers and frontend-devs - You can now open an inline browser and directly select the elements you want to edit. - The @ browser command is helpful to access and chat with your site/app and test your changes live.

100

Kevin Kern @kevinkern.dev · 03/11/2025

I see it as a simpler, more lightweight way to use Cursor in the future, while the Editor View continues to serve as the main workspace for software engineers. You can always switch back and forth with ⌘ Cmd + E

100

Kevin Kern @kevinkern.dev · 03/11/2025

2/ New Agent Window It's specifically designed to work with agents. Optimized for setups involving multiple agents like worktrees or controlling browsers.

100

Kevin Kern @kevinkern.dev · 03/11/2025

- However it likes to create a lot of files (be prepared for an army of .md files in your codebase) Composer Trivia: In the early days of cursor, the "composer" feature was the first tool that could generate and edit multiple files at once.

100

Kevin Kern @kevinkern.dev · 03/11/2025

1/ New composer-1-alpha model What I've experienced so far: - It's unbelievable FAST (similar to cheetah). - The model is tightly bound to the codebase; questions outside it are likely dropped while codebase specific info takes priority. - Works great for non coding tasks too (docs, specs, e.g.)

100

Kevin Kern @kevinkern.dev · 03/11/2025

TL;TR: - New composer-1-alpha model - Agent-focused layout (CMD/CTRL + E) - Built-in browser + Dev Tools - Parallel agents + Best-of-N selection - Voice Mode & background Plan Mode - Prompt Bar updates

100

Kevin Kern @kevinkern.dev · 03/11/2025

Cursor 2.0 update and their new agentic model "composer-1-alpha" I've been testing it for a few days and have put together a full summary in the thread below. Click to see whats new 🧵

110

Kevin Kern @kevinkern.dev · 31/10/2025

This script plays a sound when Codex finishes your task.

100

Kevin Kern @kevinkern.dev · 30/10/2025

With this, I have a small PM project in Cursor where most of the context is available to ask about the features or build all the stories and necessary tests.

000

Kevin Kern @kevinkern.dev · 30/10/2025

If you work in a larger company and your team uses Jira, I can recommend the "mcp-atlassian" MCP. It's probably the only MCP I'm currently using. I fetch all stories for a given epic in markdown, including the specification stored in confluence and export the figma screens.

100