Selçuk KUBUR

Founder of Best AI Talent Board on

[Selçuk KUBUR @Codeless33]

I made $16K in June 2026 just my salary 🥲

[Will @athcanft]

i made $13,790 in june 2026

🍎 mobile apps - $7.1K
🗣️ consults & sponsors - $2.3K
💰 X payouts - $1.7K
🔄 swaptok - $1.5K
🏭 clipfactory - $562
🐺 jarvx - $535
🔗 postbridge affiliates - $93

a lot lower than last months $21K revenue as i paused my mobile app ads mid-month - refreshed them yesterday so hopefully we smash $20K again next month

(this months profit is about the same as last month though because of the low ad spend)

[Selçuk KUBUR @Codeless33]

@damonchen @AnthropicAI always think your back pain 😀

[Unsloth AI @UnslothAI]

Qwen3.6 now runs 2x faster with MTP GGUFs! Run locally on just 18GB RAM. ⚡️

MTP enables Qwen3.6 to generate ~1.4–2.2× faster with no accuracy change.

Qwen3.6-27B MTP runs at 160 tokens/s. 35B-A3B reaches 240 t/s.

GGUFs: https://t.co/7gWhKnseZo
Guide: https://t.co/7qzk6ypWDQ

[Gregor Zunic @gregpr07]

Introducing: Browser Harness. A self-healing harness that can complete virtually any browser task. ♞

We got tired of browser frameworks restricting the LLM. So we removed the framework.

> Self-healing — edits helpers. py on the fly
> Direct CDP — one websocket to Chrome
> No framework, no rails, complete freedom
> Drop-in for Claude Code and Codex

I challenge anyone to find a task that DOESN'T work. I couldn't yet.🔥

100% open source ↓

[Selçuk KUBUR @Codeless33]

@Steve8708 Builder

[Chubby Trekka @ChubbyTrekka]

I just walked the entire length of Turkiye. I marked each camping spot/ hotel on the map. Here was my route!

[Tom Dörr @tom_doerr]

Orchestrate agents with 2,000+ skills

https://t.co/M61KrpM7eF

[Selçuk KUBUR @Codeless33]

@Dimillian Never give up 🔥🔥

[Muhib @muhibwqr]

muslims. we're 1/2 through ramadan. lock in
making du'as from the qur'an and sunnah should be simple. so i'm launching https://t.co/d9FOjyHHW8

"And to Allah belong the best names, so invoke Him by them" (Qur'an 7:180).

du'aOS uses:
> Semantic search, your intention is matched to the 99 Names of Allah and verified hadith & quran
> Hybrid ranking, vector + keyword so the right du'as surface
> AI refinement, turns that into a personalized du'a in a Prophetic style (bound by strict parameters that prevent random generation)

You get:
> Name of Allah
> Hadiths
> Quran (related)
> Your refined du'a
> Store your du'a locally in cache so you have it available as you need

No guessing. No hallucinated sources. Just intent → Name → hadith → Quran → your du'a.

check it out: 
> https://t.co/d9FOjyHHW8

this is a opensource repo, feel free to contribute and hit me up w suggestions etc.

[Selçuk KUBUR @Codeless33]

@Dimillian Push push push 🔥🔥

[Selçuk KUBUR @Codeless33]

@Dimillian When we can try 🥶🥶

[Selçuk KUBUR @Codeless33]

@Dimillian No way , can't wait to try.

[Selçuk KUBUR @Codeless33]

@sama 👀👀👀

[Selçuk KUBUR @Codeless33]

@pbteja1998 @pbteja1998 any chance to Open source this ?

[Selçuk KUBUR @Codeless33]

@Dimillian no issue at the moment will ping you on any issues.

[Ben Williams @theplgeek]

BIIIIIG 0.2.0 release for ralph-tui

Feat: Sandboxing - run all your agents in sandboxes for host FS isolation (bubblewrap on linux / WSL, sandbox-exec on mac)
Feat: New agent output parser with TUI-native rendering, more granular output and colour coding
Feat: Refactored prompt templates system with global templates and project-specific overrided.
Feat: Revised prompts that improve token usage
Feat: Added new prompt preview panel ('o' will now cycle through views - details -> output -> prompt)
Feat: Pressing 's' when stopped/idle will add +1 iterations if at max
Feat: Enabled clipboard copy in both create-prd mode and ralph loop mode
Feat: Define custom executable for an agent (enable use of agent wrappers like CC Router)
Feat: --variant flag / config option to call model with different thinking depths
Feat: Added pre-flight system to catch common ralph-tui and agent configuration issues
Feat: 'ralph-tui doctor' to detect and diagnose config issues

Fix: Incorrect detection of provider rate-limit when agent-output included @upstash/ratelimit
Fix: Skills not installing
Fix: beads tracker plugin not picking next-best task
Fix: beads tracker plugin issues not closing
Fix: Droid agent plugin not working on OS X
Fix: iteration output for completed steps will now correctly show the agent/model used for that step

Chore: Improve test coverage
Chore: Refactor to package prompt templates with tracker plugins

Some nice community contributions in here - thaanks to all.

Coming next:
- Codex agent plugin
- AmpCode agent plugin
- Gemini CLI agent plugin
- Track token usage per iteration and total in dashboard

@ryancarson @GeoffreyHuntley @danshipper

[Selçuk KUBUR @Codeless33]

@alvinsng @dabit3 Any possible integration with Codex CLI ?

[Selçuk KUBUR @Codeless33]

@saasmakermac Registered looking forward 👀👀

[Selçuk KUBUR @Codeless33]

@xenit_v0 Cloude code ve GLM ile kullanabiliyor muyuz bunu ?