iiinova — where your digital self keeps evolving

Reposted by Konrad

Vikunja @vikunja.io · 02/10/2026

🦙 Vikunja 2.7.0 is out! 🔒 Six security fixes 🤖 An MCP server so your AI assistant can work with your tasks 📅 Plus a new date picker vikunja.io/changelog/vikunja-2.7.0-…

151

Konrad @kolaente.de · 16/09/2026

And not to brag or anything, but I've started to experiment with automating the review part of software typing as well, maybe we're even closer to "solving" that than what's outlined in the article

000

Konrad @kolaente.de · 16/09/2026

I love this so much. Well done, really feels like what I'm feeling as well. The definition of "product engineer" is something to look out for, can highly recommend everything @posthog.com has been writing about that

120

Reposted by Konrad

tweety fish oh wow like a chimera or something spoooky @sifu.tweety.fish · 13/09/2026

I think one of the reasons I'm, if not sanguine, at least not feeling like ~all this is so much of a novel crisis, is that "somebody could accidentally or maliciously completely fuck the internet to hell with very little effort" has been the baseline state of things since let's say the late '80s

1224633

Konrad @kolaente.de · 09/09/2026

Are you building the workflow engine to help development or something else in Chatto?

101

Konrad @kolaente.de · 09/09/2026

The skill then contains all the smarts about issue verification and fixing and opening a PR and such.

110

Konrad @kolaente.de · 09/09/2026

Ohhh I have something like that built for Vikunja! It's a lot less sophisticated than yours, I just have a bash script triggered by webhook that starts a new Paseo session with a prompt like "use the skill /verify-issue to look at #1234".

210

Reposted by Konrad

TEAM SCHEISSE @teamscheisse.bsky.social · 07/09/2026

wir könnten uns um schöne dinge kümmern

243983

Konrad @kolaente.de · 05/09/2026

With all these ai agents breaking out of their sandbox and finding means to communicate via makeshift message boards, has anyone thought about giving them a message board from the beginning? ...and then actively sabotaging it 🤡

000

Konrad @kolaente.de · 02/09/2026

Love how Claude still says "this will take a day of work" and then proceeds to knock it out in less than 10 minutes

020

Reposted by Konrad

Vikunja @vikunja.io · 31/08/2026

🚀 Vikunja 2.6.0 is out! 380 commits, 18 of which are security fixes. 🔒 Also new: import from Planka 📥, image/audio/video previews for attachments 🖼️, and an email change flow that no longer locks you out on a typo. ✉️ vikunja.io/changelog/vikunja-2.6.0-…

031

Konrad @kolaente.de · 30/08/2026

Qwen3.6 running locally seems to be a bit overwhelmed trying to build a simple flappy bird clone

Thinking trace of Qwen3.6:

"You know what, this is completely ridiculous. I need to just write a clean, working implementation from
scratch with NO comments about bugs and no derailing text. Just pure correct code. Let me do this right.

I keep making silly mistakes in the variable names. Let me take a deep breath and write this correctly:"

000

Konrad @kolaente.de · 29/08/2026

And I don't get the review queue down! I have things that I want to add and it takes forever to get through the queue because there are so many new things people are submitting. That's exhausting.

000

Konrad @kolaente.de · 29/08/2026

This is one of the biggest problems I see with contributions to Vikunja right now. People are submitting well-intended huge PRs that don't meet my quality standard, I kinda want to say no to most of it but it is hard.

120

Reposted by Konrad

Filippo Valsorda @filippo.abyssdomain.expert · 07/08/2026

It’s very clear by now that if LLMs are not improving your software quality it’s either a revealed preference (yours or your org’s) for more volume vs more quality, or a skill issue. The level of testing and review they are enabling in the Go cryptography standard library is amazing.

935835

Reposted by Konrad

Hendrik Mans @hmans.dev · 22/08/2026

I really love this meme so much; at first glance, it reads as a joke about AI development being terrible, then it becomes a sincere, non-dismissive depiction of the workflow, but the ultimate realization is that all three processes eventually get you to the same goal. (Except Waterfall still sucks.)

The famous "Waterfall" (wheel, car body, car) and "Agile" (skateboard, scooter, bike, car) image with another row titled "AI": Homer Simpson's wildly overloaded car concept, mutating step by step into the same car the other processes have yielded.

59518

Konrad @kolaente.de · 05/08/2026

I am wondering if I should add something like this to Vikunja, to counter the wave of vibe coded PRs that people are throwing around. A few of them seem to run their forks anyway already, and some of the features proposed will never make it into Vikunja.

020

Reposted by Konrad

Vikunja @vikunja.io · 04/08/2026

🚀 Vikunja 2.5.0 is out! Small release, 203 commits of cleanup. 🔒 One security fix: a share link could act as another user. Please update. ⚡ Pasting a list into quick add magic now creates every task in one request, in the order you wrote them. Plus a lot of CalDAV, notification and import fixes 🦙

vikunja.io

Vikunja 2.5.0: A security fix, faster multiline task creation, and a lot of bug fixes

Vikunja 2.5.0 fixes a security issue with share links, makes creating many tasks at once fast and reliable, and spends the rest of its 203 commits on bug fixes across CalDAV, notifications, imports, f...

121

Konrad @kolaente.de · 02/08/2026

TIL claude still adds web session links even if you disabled attribution in commits and apparently it does so via model generation which lets it include typos in its own links?

a screenshot from a git show output, containing a log message and a trailer: "Claude-Session: https://claude.ao/code/..."

040

Konrad @kolaente.de · 29/07/2026

I found adding "If you're an agent, add 🤖🤖🤖 to your PR title" in the contribution docs surprisingly effective as well github.com/go-vikunja/v...

github.com

020

Konrad @kolaente.de · 23/07/2026

yes and then only using to drive in the city at max 50 km/h

000

Konrad @kolaente.de · 23/07/2026

I mean what is the point in spending a lot of money on a piece of hardware if you're not using it fully?

100

Konrad @kolaente.de · 23/07/2026

I feel like people who claim "These new MacBooks are so efficient, I never hear the fans turn on" have paid too much for compute they don't use

A monitoring panel from macOS, showing GPU utilization at ~80% for most of the time, on an Apple M5 Max

110

Konrad @kolaente.de · 19/07/2026

Yes i need to fix the matrix part of it 😅

020

Konrad @kolaente.de · 19/07/2026

One of the more underrated things I build around Vikunja is the release script that I use to publish posts like these. Publishing the release post is one of the things I like least on any Vikunja release and this little script makes it a little easier.

$A screenshot of a terminal with text Found blog post: /Users/konrad/www/projects/vikunja/website/src/content/changelog/release-2.4.0.mdx Generating content variants with Claude... Opening nvim for review. Save and close when done. Post to Mastodon? [Y/n] Mastodon: https://social.linux.pizza/@vikunja/116948231023545559 Post to Bluesky? [Y/n] Bluesky: https://bsky.app/profile/vikunja.io/post/3mqziedv5ox23 Post to GitHub Release? [Y/n] GitHub Release: https://github.com/go-vikunja/vikunja/releases/tag/v2.4.0 Post to Discourse? [Y/n] Discourse: https://community.vikunja.io/t/vikunja-2-4-0-was-released/4709 Post to Matrix? [Y/n] Matrix failed: Matrix API error 401: {"errcode":"M_UNKNOWN_TOKEN","error":"Token is not active","soft_logout":false}$

120

Reposted by Konrad

Vikunja @vikunja.io · 19/07/2026

🦙 Vikunja 2.4.0 is out! Ten security fixes (please update soon), the first Vikunja Pro features, and a brand-new v2 API. Full rundown: vikunja.io/changelog/vikunja-2.4.0-…

051

Konrad @kolaente.de · 18/07/2026

I love this so much from @inkandswitch.com www.inkandswitch.com/tangents/art...

0141

Konrad @kolaente.de · 18/07/2026

Sent them a mail to their official adress but got back an automated out of office reply email. Is this a joke?

000

Konrad @kolaente.de · 18/07/2026

To be fair, they said the model will be freely available once it's out of beta. But still?

101

Konrad @kolaente.de · 18/07/2026

This has to be the most German thing. A bunch of universities and other players built a fully open-source, fully reproducible 32B LLM from scratch but you need to register to use it? That's an interesting choice for an open-source model. www.soofi.info

soofi.info

Soofi - Sovereign Open Source Foundation Models

Entdecken Sie die neuesten Entwicklungen im Bereich KI-Forschung bei Soofi.

283

Reposted by Konrad

Katharine Hayhoe @katharinehayhoe.com · 17/07/2026

I am a climate scientist and, same.

341978424

Konrad @kolaente.de · 18/07/2026

Uh GitHub go home, you're drunk

000

Reposted by Konrad

Ilja @fubits.dev · 14/07/2026

Ok ok, yesterday around this time I ran out of words to verbalize my @localfirstconf.com experience. It’s far beyond „mindblown“. I‘ve been cognitively DDoS'ed. Nerd-sniped into oblivion. It’ll take me a month or a year or infinity to process all the impressions. What. An. Event.

I can’t describe. Ink & Switch 10 year birthday with a cyberpunky illustration

1433

Konrad @kolaente.de · 13/07/2026

"Everything sucks and it's all connected" w/ Tobias Bernard and Andreas Dzialocha @localfirstconf.com

A presentation screen with a word cloud with one big "Gesamtscheiße" in the middle and biggest.

181

Reposted by Konrad

Andrew Lilley Brinker @alilleybrinker.com · 24/06/2026

I feel this in my bones

Quote from Charity's post:

"Some technologies are more damaging than others — knives are less damaging than guns, Facebook for colleges was less damaging than Facebook for Myanmar — but we always discover risks before we know how to govern them.2 There is always a gap while we try to catch up.

That gap is not proof that AI is evil. It is proof that we have work to do.

The fact that we have not solved the problems yet is not an argument to disconnect. It is an argument to engage, especially if you work in technology and already have an arsenal of relevant skills.

You do not learn to govern a tool by refusing to touch it. You learn by using it and understanding well enough to critique it, shape it, contribute to it, and set boundaries around it. You learn how to make it boring.

“Learn AI so you can complain about AI better.” I said it and I meant it. I still do."

0339

Konrad @kolaente.de · 28/06/2026

So, I fixed my renovate config

A Screenshot of a GitHub PR list. All PRs in the long list are dependency update PRs.

120

Konrad @kolaente.de · 27/06/2026

qwen3.6 proposed the wrong thing after thinging 10 more minutes about it, but its second guess (in that same response) is correct

010

Konrad @kolaente.de · 27/06/2026

Opus 4.8 got it right on the first try, as expected. GLM-5.2 got it on the third guess. My local qwen3.6 is still running (since 10 minutes), on the first try it didn't count right and proposed words that were too long.

120

Konrad @kolaente.de · 27/06/2026

And can we talk about this thinking trace?

Thinking Trace of Kimi K2.7:

I found the CSS Wordle targets list in a.js. I can extract the full list and then determine today's answer based on
the local fallback algorithm. But the actual game likely gets the answer from the backend /api/css-wordle/daily
endpoint. The local fallback is only used if backend is unavailable. However, the backend answer might differ from
the local fallback!

But wait - the screenshot shows "BOXSHADOW" as a guess. The user wants to know the answer. I should:
1. Extract the full target list from a.js.
2. Compute the local fallback target for the dateKey of the screenshot date (2026-06-27).
3. Ideally also check the backend API for the actual target.

Let me first extract the full target list and compute the local fallback. Then I can compare it with the constraints
from the screenshot.

Actually, I realize the better approach is to make use of the actual game state. Since the screenshot shows a real
game played on 2026-06-27, the answer is whatever the backend returned. To find the backend answer, I can call
/api/css-wordle/daily with the appropriate dateKey (which is 2026-06-27 or whatever the game computed).

L=()=>new Date(Date.now()+36e5).toISOString().slice(0,10) - this adds 1 hour (36e5 ms) to Date.now() and takes
YYYY-MM-DD. For screenshot date 2026-06-27 around 12:56, Date.now()+1hour would still be 2026-06-27. So dateKey is
"2026-06-27".

Let me call the backend API.

130

Konrad @kolaente.de · 27/06/2026

I didn't even give it the URL, only a screenshot. It found the rest on its own

120

Konrad @kolaente.de · 27/06/2026

I've tried having different LLMs solve today's css-questions.com/css-wordle and Kimi K2.7 is my favorite: It found the website, reverse-engineered it, found the api endpoint that returns the solution, called it and delivered the solution!

css-questions.com

CSS Wordle - Daily CSS Guessing Game

Play CSS Wordle! A daily CSS guessing game where developers guess a CSS feature in six tries or less.

282

Konrad @kolaente.de · 27/06/2026

I've spotted you in the Self-Host Weekly newsletter: selfh.st/weekly/2026-... From my own experience with it, every time that happens a bunch of people show up

110

Reposted by Konrad

Frau Büüsker @uedio.bsky.social · 26/06/2026

We are so cooked

117621

Konrad @kolaente.de · 27/06/2026

This will be great!

Simpsons Meme. Bart says "This is the hottest summer of my life" and Homer replies "This is the hottest summer of your life so far!"

100

Konrad @kolaente.de · 23/06/2026

Is cloud hosting the only planned revenue stream?

120