IG: @Brewster.a

Am me

[José Siles | AI | Data @josesilesdata]

DESCANSE EN PAZ VIBE CODING:

2025 - 2026 🪦

GitHub ha sacado spec-kit

La idea: dejas de pedirle cosas al agente a lo loco y escribes primero la especificación.

Luego el agente la ejecuta.

El flujo son 5 comandos:

/speckit.constitution → los principios de tu proyecto
/speckit.specify → qué quieres construir
/speckit.plan → stack y arquitectura
/speckit.tasks → lista de tareas real
/speckit.implement → lo construye

Funciona con 30+ agentes:

- Copilot,
- Claude
- Codex
- Cursor

y muuuuchos más.

+ 125,000 ⭐ en GitHub

Link al repo en los comentarios👇🏼

[Venkatesh @venkateshdotdev]

Someone found our public signup endpoint and scripted it.

No rate limit.
No CAPTCHA.
No email verification gate.

40,000 fake accounts in six hours.
Support thought it was a viral moment.

It was a bot with a for loop.

How do you rate-limit signup without punishing real users?

[Vox @Voxyz_ai]

𝘀𝘁𝗼𝗽 telling Claude Code/Codex "no em dashes".
𝘀𝘁𝗼𝗽 telling Claude Code/Codex "stop saying delve".
𝘀𝘁𝗼𝗽 telling Claude Code/Codex "don't sound like AI".

you never gave it a 𝘄𝗿𝗶𝘁𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺. every README, PR description and landing page ships in the same AI voice, and you ban words one at a time.

orwell wrote the fix 𝟴𝟬 𝘆𝗲𝗮𝗿𝘀 𝗮𝗴𝗼. six rules, 1946. paste them into your global CLAUDE.md / AGENTS.md and every session picks them up:

→ never use a long word where a short one will do
→ if it is possible to cut a word out, cut it out
→ never use the passive where you can use the active

6 blocks you can copy-paste directly 👇

[Neo Kim @systemdesignone]

If you want to become god-level with AI agents, save these 10 GitHub repos:

[MIKE @mikenevermiss]

Claude Fable 5 is actually a beast for web design.

this guy just dropped a full guide on how to design and build interactive animated 3D websites with Claude Code + Fable 5

(complete prompts in the comments)

watch the video and read the full guidelines on the article below

[Meng To @MengTo]

This is a great way to learn UI prompting. Lots of examples with interactive demos.

Also check out https://t.co/9yTZpa9yjm and https://t.co/3HRT8RycUw. Knowing UI patterns gives you the vocabulary to describe interfaces with taste.

[Khairallah AL-Awady @eng_khairallah1]

Anthropic engineer:

"You're not supposed to prompt Claude. You're supposed to build a system that prompts itself."

In 45 minutes she shows exactly how to build an agent that improves itself.

Most people are still doing all of this by hand.

Watch the session, then save the guide below.

[Karan @karankendre]

Claude charges for image inputs based on pixel size, not the amount of text inside the image

So you can literally turn long prompts into an image and save a lot of tokens

Here is a repo which does this

[IG: @Brewster.a @brewster254]

😂ati Jamba nn?!

[Masu Zafi 🔥🔥 @masuzafi]

🚨 The gods must be CRAZY.
He probably told his followers he could fly.
A 49-year-old Hindu priest in Andhra Pradesh gathered everyone on a mountain top.
He said he had divine powers.
Performed rituals to please the god “Jamba Malaya”…Then walked to the very edge with total confidence.

[IG: @Brewster.a @brewster254]

@Manu_Sisti Claude

[Men of Purpose @Men_Of_Purpose]

She literally explained what to do if you’re not financially where you want to be.

[Oliver Prompts @oliviscusAI]

🚨 BREAKING: Someone just open-sourced software that sees you through walls using only WIFI signals.

it’s called WiFi-DensePose. It maps your exact body pose in real-time. no cameras. no sensors. just your living room router.

100% Open Source.

[IG: @Brewster.a @brewster254]

@aashatwt Link

[IG: @Brewster.a @brewster254]

@matt_elevenlabs 11B

[Omah🌺 @TassyOmah_]

Another one is YC website.

Lots of new startups always without a design team and structure.

Check their startup idea, emphasize your value, Write to their founders on directly YC website, LinkedIn, email etc

Out of 20, chances are five will respond to you.

[darren @darrenjr]

i still want an ai competitor watcher

just 24/7 monitor my competitors - blog posts, rss feeds, git commits, tweets, linkedin, openapi spec updates etc

summarize it and alert me in slack

i’d pay for this right now

[GREG ISENBERG @gregisenberg]

30 ways to find your next $10K+ MRR idea for 2026:

1. Read GitHub issues and look for recurring pain points developers ignore.

2. Set Reddit alerts for “I wish someone would build…” and validate demand.

3. Build an agent around a single recurring vertical Upwork task that pays well.

4. Monitor API changelogs and build integrations the day they launch.

5. Summarize 1-star Chrome Store reviews with ChatGPT and fix the top three complaints.

6. Audit browser DevTools to see what power users still do manually.

7. Reverse-engineer topProduct Hunt hits, then apply AI to improve them.

8. Read YouTube tutorial comments to see what viewers still can’t figure out.

9. Watch Twitch streamers and note what workflows interrupt their flow.

10. Scan job listings for repeated “must-know” tools; build easier versions.

11. Dig through graveyard from companies like Google and ship the best abandoned projects.

12. Implement new AI research papers as usable web apps.

13. Explore niche subreddits and find problems that appear every week.

14. Review SaaS feature requests and build what the big players delay.

15. Connect open-source tools that don’t yet talk to each other.

16. Track “Chrome extension for X” search volume to spot new demand.

17. Feed GPT the top extension descriptions and ask for adjacent product ideas.

18. Use Perplexity Deep research etc to mine podcast transcripts on people's daily frustrations they’re telling you what to build.

19. Follow changelogs and tech-stack migrations of popular startups; build the missing glue.

20. Look at Zapier’s most-used zaps and each one could be an autonomous agent.

21. Track “AI for X” or “agent for X” search queries with SerpAPI.

22. Analyze public Notion templates and build vertical agents around them.

23. Browse LinkedIn for people describing manual data tasks and productize one.

24. Watch how startups use ChatGPT for customer support and make a vertical agent from it.

25. Rebuild niche directories (lawyers, therapists, realtors) as AI concierge services.

26. Create agents that plug into boring SaaS categories: procurement, compliance, HR ops.

27. Read changelogs from AI model providers; build tooling the day new capabilities appear.

28. Find spreadsheets that companies rely on and replace them with AI dashboards.

29. Identify agencies that charge per project and productize their work into recurring SaaS with agents.

30. I built a tool that automates a lot of this, https://t.co/a5ARFnvky2 and we give away 1 free startup idea per day with paid plans for AI agents to help you. Maybe it'll get your creative juices flowing.

I'm rooting for you.

TLDR;

– The next big idea is probably hidden in a comment thread.
– Every complaint online is a free focus group.
– Chase friction. every pain point is a map. find one, solve it, then look for the next one it connects to. keep solving until the solutions form a workflow people can’t live without. by owning the whole chain of pain you build defensibility.
– Every repetitive task is a business model waiting for an agent.
– The internet keeps leaving clues, just gotta listen

[Prajwal Tomar @PrajwalTomar_]

Most people spend days designing mobile apps.

I designed an entire mobile app in 20 minutes using AI.

Everyone’s talking about vibe coding.

This is vibe designing ↓

[Jacob Boar @JacobDzikowski]

Here's how to make $10k MRR with B2B whop apps using their new SDK:

> Go to whop documentation
> Get(scrape) all of their documentation and save to PDF
> Go to Chatgpt add the PDF and send this prompt:

TASK:
You are provided with the information about 'WHOP apps', the new opportunity to make money with building software.

Your role is to find the golden idea for a WHOP app that has potential for at least $10k MRR.

In order to do that think deeply and perform analysis following these steps:

1. Analyse provided document to understand what are whop apps and all the endpoints whop SDK is offering to understand the possibilities and constrains of building these apps.
2. Analyse user context presented as a JSON object to understand user's unique positioning, experience, niche.
3. Find 10 unique ideas and present it to the user following the output format instructions.

USER CONTEXT:
<drop a context profile here describing your experience, niche etc. or use plain english to do that>

OUTPUT FORMAT:
Your output should be a numbered list of 10 personalised ideas(1-10) with the following parameters:

1. Idea desciption
- Pain: <describe what pain does this app solve>
- Dream outcome: <describe the desirable state of the user WHY he would even bother spending his valuable time using this app>
- Vehicle: <how does the app transition the user from pain to dream outcome, describe the core functionality of the app>
- ICP: <which group most benefits from using this app>
- Distribution: <where to find the congregation of our icp and how to target them to promote the app>
- Why: <what makes this idea good fit for the user to build>
2 ...

FINAL NOTES:
- ALWAYS ground your answers on the provided context
- ALWAYS think from the user perspective "why would I even bother using this app?!"
- ALWAYS maintain brutal honesty when talking with the user and providing ideas pushing him to validate that the pain points are real, frequent and monetizable