Yanko Enchev

[Scott Stedman @ScottMStedman]

Russian code is inside the White House app, which is now mandated on all government phones. The company behind the code is growing. Its founders bank at sanctioned Russian banks. And its accountant has a criminal history. My investigation:

[Zhengyao Jiang @zhengyaojiang]

The first experimental evidence of recursive self-improvement (RSI).

Autoresearching the autoresearch agent for eight days.

The result beats the harness we hand-tuned for two years, on held-out benchmarks: 🧵(1/7)

[Bryan Ng @boomerrbryan]

There's a play printing money right now that's so simple it'll piss you off

People are taking 60-second TikToks that went viral, finding the TOPIC behind them, and rebuilding that topic as a 12-minute YouTube video where the same idea pays 50x more

A TikTok about "why Japan's economy makes no sense" gets 4 million views and earns the poster maybe $40 from the creator fund. Mid

The same topic rebuilt as a 12-minute YouTube deep dive with an AI character earns $15-45 per thousand views. 200k views on YouTube at $30 RPM = $6,000. From a topic that was already proven to work. On a platform that pays real money

The TikTok did all the validation for you. 4 million people already told you they care about this topic. You just moved it to a platform where attention is worth something

The workflow is almost too dumb to type out:

Scroll TikTok for 30 minutes. Screenshot every video over 1M views that covers a topic you could expand into 10+ minutes. History, money, science, geopolitics, psychology, anything with depth behind it

Take the topic, not the video. Research it for an hour. Script it with AI. Have your character deliver it. Upload

The TikTok audience already proved demand. You supply the long version on a platform that actually pays

People treat TikTok and YouTube like competitors. They're not. TikTok is a free focus group that tells you exactly what millions of people want to learn about. YouTube is where you get paid to teach it

Most viral TikToks are 60-second summaries of topics that could easily fill 15 minutes. The short version gets views. The long version gets revenue. And nobody is connecting the two because the people on TikTok don't think about YouTube and the people on YouTube don't scroll TikTok

One platform tells you what to make. The other one pays you for making it

(we currently have a waitlist for the tool that replaces your entire production team and makes YouTube actually work for your business. waitlist link is in bio. spots are limited.)

[Everintrigued @EverIntrigued]

Russia trains their sex spies to give the PERFECT blowjob 😭

Deep throat technique, banana practice, 90 degree head angle mandatory.
Your husband will thank you (or hate you).
This podcast went too far and I’m here for it 😂

[Miles Deutscher @milesdeutscher]

I just published my masterclass on how to get rich with AI.

No hype or gimmicks - just realistic money-making methods that anyone can get started with today (based on my real experience).

Easily one of my best videos of the year.

Watch now 👉 https://t.co/q0b7Qrwy17

[Nick @nickbaumann_]

one more tip: have the "babysitter" maintain a visual progress update in the in-app browser (@`browser`)

[Nick @nickbaumann_]

Pro tip: when prompting Codex with really difficult /goals, ask it to "write a goal for another thread to achieve this and babysit it until it figures it out"

By doing so, you'll add built-in steering and another layer of taste verification (that's how this video was made)

[Alex Prompter @alex_prompter]

Someone distilled Fable's problem-solving approach into a plugin any model can run. 159 agent runs later, Sonnet + the plugin matched Fable 10/10 on a research task.

The plugin has three parts. fable-method gives any model a structured problem-solving loop with hard failure thresholds. fable-loop runs entire tasks with adversarial verification agents checking the work. fable-judge treats every "done, all tests pass" claim as unverified and re-runs everything independently.

The results are specific. Haiku went from 0/4 to 4/4 on catching a wrong test before "fixing" correct code. Sonnet + the method matched Fable itself on a five-part research task, scoring 10/10. The judge took Haiku from 3/5 to 5/5 on catching planted frauds in a fake completion report.

The part that earned my trust is what it won't do. On ordinary tasks with capable models, it adds nothing. They measured that, documented it, and the README says so. The value concentrates at traps like wrong tests, false completion claims, and weak models running unattended. It supplies discipline, not knowledge.

The iteration tells you a lot about the quality. v1 of the method failed its own test at 0/4. v2 failed at 1/4. v3 finally passed at 4/4. The repo includes every failed version with raw judge transcripts. The flowcharts weren't built from Fable describing how it thinks either. They recorded actual tool calls from bare Fable agents and corrected the method wherever Fable's self-description disagreed with its real behavior. Observation beat self-reporting 3 times.

If you're running Claude Code with cheaper models or want to harden your agent workflows, this is worth 10 minutes of your time.

To install in Claude Code, run /plugin marketplace add Sahir619/fable-method then /plugin install fable@fable-method

[Kappaemme @Kappaemme1926]

CODEX SKILL THAT FINDS YOUR STARTUP’S FIRST CUSTOMERS!

I made a Codex skill that analyzes your startup and finds potential customers from real public signals.

Paste your startup URL while Codex defines your ideal customer, searches public discussions, qualifies each prospect, and generates a polished report with personalized outreach openers.

-> ideal customer profile analysis
-> public pain + buying signal research
-> evidence-backed prospect shortlist
-> fit, timing + reachability scores
-> original source links for every prospect
-> personalized outreach openers
-> polished HTML report
-> one-command install

Install: npx --yes codex-first-customer-finder-skill

100% open source.
Repo in Bio.

[Marcos Hernanz @MarcosHernanz]

After doing ~50B tokens, this is the config I've landed on

[Sudo su @sudoingX]

to everyone who runs qwen 3.6 27b dense, what hardware did you land on, what's your top speed, and what's the sweet spot context before it starts dragging?

any hardware, any quant. tok/s, usable window, asking for fren.

[Marcos Hernanz @MarcosHernanz]

The best Codex config

[Justin Brooke ❤️‍🔥 @IMJustinBrooke]

SERIOUS QUESTION…

Literally what is stopping some of you from making money online right now?

With Fable 5 and tools like Replit or Lovable, I can’t fathom how it’s possible.

Please comment and tell me. I’ll help you for free, 20yrs full time online. Help me, help you 💪❤️

[Kai @hqmank]

GPT-5 models are chatty by default.

Ask it to fix one line. It explains the problem, your code structure, the fix, and why it made that choice.

Set verbosity: "low" in your config and it just gives you the line.

Faster. Cheaper. Actually usable in agent workflows.

[Vaibhav (VB) Srivastav @reach_vb]

codex tip: ask codex to do its research first and then use set_goal to set an appropriate goal instead of /goal

It results in massively better prompt and downstream results!

all my prompts would start with requirements gathering and research and once done set_goal

[Jayden ⛩️ @thejayden]

I GENUINELY DON’T UNDERSTAND WHY EVERYONE ISN’T USING THIS YET.

The CEO of Obsidian just open-sourced his Claude Code skills.

You literally just tell Claude to clone this repo and start using them.

https://t.co/kWfVGQseM5

[Hugging Models @HuggingModels]

Big news for AI on a budget. GLM 5.2 Colibri int4 is a Mixture of Experts model that runs entirely on your CPU. No GPU needed. It's fast, efficient, and opens up advanced language capabilities to anyone with a standard computer. This is a game changer for offline AI.

[Haider. @haider1]

GPT-5.6 Luna is a standout model for price-to-performance

at half the cost, "5.6 luna max" matches gpt-5.5 xhigh, while outperforming "fable 5 md" at half the cost -- so together, "luna" and "sol" dominate the Pareto frontier

i'm still unsure when to use "terra"

[0xSero @0xSero]

GPT-5.6-Sol (medium) is the optimal configuration.

[Miraqle @0xMiraqle]

Claude can't, but GPT 5.6 on GOD-MODE is IMMACULATE

Here's how you create it step by step:

> open the new desktop app, pick Sol, reasoning on High. this is taste work, don't give it to the small models

> feed it 2-3 sites you love and one line: "extract the art direction: mood, motion, typography, pacing. write it down as a style bible"

> then the brief, one paragraph, goal not steps: "resort site for [name]. cinematic scroll, the booking button always one glance away. follow the style bible"

> add house rules: no template hero-with-three-cards, no stock gradients, motion carries the story, every section earns its scroll

> set the bar: "a working designer can't tell this from an agency build." then spin up a SECOND 5.6 with fresh context to grade against that bar. the builder never grades itself

> loop it: build, grade, close the biggest gap, again. walk away, it doesn't need you in the room

> when the verifier runs out of complaints: tag Sites. live URL, one click, no hosting, no deploy

The deeper version of every step (the contract, the rules, the verifier trick, when to spend on Ultra) is in the article below.