🏔️ Peak

2.3K posts

🏔️ Peak banner
🏔️ Peak

🏔️ Peak

@peaklabs_dev

👨‍💻 Full-Time Entrepreneur & 👨‍🍳 Core Developer at coolLabs on @coolifyio. 🚧 Building OSS at https://t.co/91ji7n8yrq

🏔️ Peak One Katılım Ocak 2024
719 Takip Edilen1.6K Takipçiler
Sabitlenmiş Tweet
🏔️ Peak
🏔️ Peak@peaklabs_dev·
🚀 I am excited to announce that I am now officially a Core Developer at @coolifyio! 🎉 Coolify has become my passion and I'm committed to taking it further than ever imagined. I am excited to shape the future of Coolify together with the community.
Andras Bacsai@heyandras

It is Monday and I have exciting news! 🎉 Over the past weeks/months, I realized that @coolifyio is growing on me. 😅 So thanks to the support of the community, sponsors, and cloud users, Coolify now have its first paid core contributor. It's unbelievable, right? It certainly feels that way to me! Please join me in welcoming @peaklabs_dev! 🎉 What's even more incredible is that I wasn't even searching for someone. He simply began contributing useful PRs for free and showed me how he can do things. Nice. This is the way.

English
5
5
128
32K
Bartek Michalak
Bartek Michalak@barmic12·
@ClementDelangue We recently added a short instruction for agents that at least helps catch people who won't even bother to glance at the PR their agent created 🥞
Bartek Michalak tweet media
English
2
2
38
2.2K
clem 🤗
clem 🤗@ClementDelangue·
Our biggest open-source repos are getting overwhelmed by AI slop which literally makes Github unusable (~a new pull request every 3 minutes). Fun new challenges in an agentic world!
clem 🤗 tweet media
English
160
106
1.3K
189.7K
kero
kero@Kerollmops·
@ClementDelangue We've got the same recently on @meilisearch... we introduced a new no-agent AGENTS.md to ensure we are not spammed too much!
kero tweet media
English
1
0
2
707
Pushpak
Pushpak@pushpak1300·
Laravel has now cute little ai policy for community contributions. 🥰
Pushpak tweet media
English
4
10
89
6.4K
🏔️ Peak
🏔️ Peak@peaklabs_dev·
2026 Development Log 45 & 46 🚀 Today I absolutely nerd-sniped myself into the deep details of how SQLite and Postgres store each different column, how many bytes they use and more. 👨‍🍳 Core Coolify Development ✅ Continued work on the v5 database schema ✅ Built a PHP package which does some clever tricks to make ULID performance as good as UUIDs when using huge tables (not sure if the package is prod-worthy yet) ✅ Improved the sidebar design a lot - reordered menus, improved consistency and improved notification page design to allow for app-level notifications and more ✅ Nerd-sniped myself deep into SQLite vs. Postgres vs. MySQL for column type storage, bytes, UUIDs... ✅ Did some more digging and testing for ID, UUID and ULID ✅ Updated 2 v5 plans ✅ Added 1 new plan for storages ✅ Provided some Discord support 🩹 Health ❌ Skipped my walks
English
1
1
13
644
Tobias_Petry.sql
Tobias_Petry.sql@tobias_petry·
I feel honestly sorry for any maintainer of big open source projects. So many slop PRs they are now bombed with and have to close... And you see this from many authors. Always many PRs created one after another. They often do this multiple days until giving up.
Tobias_Petry.sql tweet media
English
11
5
59
12.4K
🏔️ Peak
🏔️ Peak@peaklabs_dev·
@ahlimanhuseynov ULID with maybe some tricks to make it as performant as UUID. I'll probably write a more detailed post soon.
English
0
0
2
16
AHU
AHU@ahlimanhuseynov·
@peaklabs_dev I am just curious, what did you choice between UUID and ULID and why?
English
1
0
1
33
🏔️ Peak retweetledi
🏔️ Peak
🏔️ Peak@peaklabs_dev·
2026 Development Log 43 & 44 🚀 👨‍🍳 Core Coolify Development ✅ Started building out 70% of the database schema and models with the new rules, naming and more ✅ Did a deep dive and made decisions about new primary keys, IDs, UUIDs, ULIDs, slug usage and their migratability from v4 ✅ Created and finished v6 of the v5 sidebar design and added an integration plan page for it ✅ Permissions for v5 were planned out a while ago but scopes for scoping to an app for example, not really. I do not want to be happy too early but I think I finally figured out a clean way to do scopes. ✅ Added 2 new plan pages to the refactoring guides one about API refactoring, the other about URLs, primary keys and slugs ✅ Updated 4 v5 plans ✅ Made a lot of decisions about naming and other questions that were still open 🩹 Health ✅ Went for 1 short walk
English
1
1
17
1.1K
wukko
wukko@uwukko·
we now get at least one ai slop PR per day, this is an open source pandemic
wukko@uwukko

@aritmiabattito the amount of slop prs is tolerable for now, so not yet at least

English
12
11
566
20.9K
🏔️ Peak
🏔️ Peak@peaklabs_dev·
@MathieuJ85 Not sure how deep the tests will go yet but a loot of tests will be added for sure.
English
0
0
2
15
Mathieu
Mathieu@MathieuJ85·
@peaklabs_dev Would be greaaaat if in the ci there could be an e2e test that create a project, add a mongodb database, copy db connexion, add private git repo, set env, add domain and deploy and then when finished test if app is deployed on domain. Then clean the test server by removing each
English
2
0
1
78
🏔️ Peak
🏔️ Peak@peaklabs_dev·
2026 Development Log 42 🚀 Today was a very successful day as I merged 3 PRs for v5 and finally our comprehensive v5 tests workflow is running all green (see image). 👨‍🍳 Core Coolify Development ✅ Finished JS tooling PR and merged it ✅ Opened 1 new PR with a much improved tests workflow using all PHP and JS tooling and merged it ✅ Rebased, solved like 20 merge conflicts and then improved the upgrade Inertia v2 to v3 PR and then merged it ✅ Started work on the v5 part of the v4 to v5 upgrade migration ✅ Planned clean Svelte component and page naming conventions (need later next week) ✅ Updated internal commit, PR, Laravel and other coding and quality rules 🩹 Health ❌ Skipped my walk
🏔️ Peak tweet media
English
1
1
11
1.6K
John
John@DataWhispurr·
Stop AI-generated code from polluting your repo. anti-slop-skill — purpose-built Claude Code skill that scans for: • Hallucination signals & hedging • Security hazards & leaked keys • Cross-language patterns in Python • Dead noise, placeholders, over-abstraction Fresh subagent judges its own output. Configurable thresholds. Git hooks + Actions ready.
John tweet media
English
1
0
5
97
🏔️ Peak
🏔️ Peak@peaklabs_dev·
Mom I made it! My project is in a @theo video 🥳. Just to clarify some things. ◆ Anti slop does not use AI to catch AI so it is essentially free, it runs in 15s at no cost. ◆ The checks are working very well on @coolifyio (51K stars), it already successfully caught 199 slop PRs in the last 30 days. ◆ I am aware that there are many options but that makes it flexible and able to catch slop in different repos and allows for configurable sensitivity so that new contributors are not flagged just because they are new.
English
19
17
877
84.2K
niji
niji@nijikokun·
@peaklabs_dev @theo @coolifyio Oh, I misunderstood the meaning of it, this is actually reasonable either way, I hate dealing with half of these.
English
1
0
1
27
🏔️ Peak
🏔️ Peak@peaklabs_dev·
I don't distinguish between human created AI slop and machine created AI slop because they're both slop. If a human writes something very template-y, it should still pass the checks because most checks are built in a way that a human writing a PR description, a change ... will pass it.
English
0
0
2
161
Noah
Noah@Noahhh1005·
@peaklabs_dev @theo @coolifyio 199 catches in 30 days on a single repo is a wild number. curious about the heuristics—how do you distinguish between "AI wrote this PR poorly" vs "human who just writes very structured, template-y commit messages"? feels like the sensitivity tuning is where the real art is
English
1
0
1
473