https://x.com/basedonex/status/2042546883411779841?s=46

PSA: CLAUDE MAX / CHATGPT PRO REBATES

If you are not a Singapore citizen/from any of the non-excluded country list, you can use the based card to get free ChatGPT/Claude Max sub (up to $100usd) rebated on the based gold card.

Good read, especially with all the AI stimuli.

Spoke to some friends who feel they are very left behind because of how fast things are. The reality is still we are in an echo chamber. The masses still have not caught on yet. We think we are lagging behind because of the algo we are being fed.

Modern life feels dull because our brains are getting trained by “superstimuli” - social media, hyper-edited content, instant novelty, vanity metrics, junk food, etc. These hit our reward systems harder than normal reality, so ordinary life starts to feel flat.

The key point: the issue isn’t just “too much dopamine,” but that our baseline expectation for stimulation gets pushed so high that real life can’t compete.

TLDR:
If you want life to feel more real again:
- reduce engineered stimulation
- lower your baseline
- spend more time in lower-stimulation states
- stop judging real life against optimized digital experiences

https://x.com/thedarshakrana/status/2042643388642791893?s=46

If you are running Claude code, try this /advisor skill. It’ll save you a lot of tokens.

Pair Opus as an advisor with Sonnet or Haiku as an executor, and get near Opus-level intelligence in your agents at a fraction of the cost

China’s Ministry of Education has launched a comprehensive initiative to integrate AI throughout the national education system, spanning from K-12 to university levels.

Under this plan, AI will become a mandatory public course for college students, complemented by cross-disciplinary curricula at all grade levels.

The Ministry aims to establish a fully realized AI education framework by 2030.

Claude Opus has been nerfed. recently hallucination score have put it its accuracy from 83.3% in the past to 68.3% (in april)

I heard a fix for this is use “/effort max” everytime you start a new session, otherwise it’ll just auto reset to “high” or edit your settings.json

Helped a couple troubleshoot their OpenClaw setup today. Nothing beats seeing those happy, excited faces when they birth their Openclaw 🦞🦞

The first pilot gave us some great data on the friction points, so we've refined the whole workflow for the next round.

Catch the next workshop on
28 April if you need some guidance to create your first AI employee!

https://luma.com/htvx3vuo

Save this list down and install these skills for Claude to help you cut down on token usage
👇👇👇

1) Plugins / Skills (extend what Claude can do)
👉 https://github.com/obra/superpowers-lab

Based off the OG superpowers skills, this one adds more experimental features to also help you cut down on token usage

2) Memory (Claude remembers things for you)
👉 https://github.com/thedotmack/claude-mem

Think of this like giving Claude a brain that doesn’t reset every time.
Instead of re-explaining your project, style, or preferences every session… it remembers.

⸻

3) Context Manager (only show Claude what matters)
👉 https://github.com/Sceleratis/claude-code-stuff/tree/main/plugins/context-manager

Right now, most people dump everything into Claude. This tool filters and feeds only the relevant parts.

⸻

4) Smart Search (Claude can “look up” your code instead of loading all of it)
👉 https://github.com/danielbowne/claude-context

Rather than loading your whole project into memory, Claude searches and pulls only what’s relevant.

Finally wrote something long-form

First Substack post is up - ParentOS.
It's about using agentic AI for young families: inbox aggregators that catch school logistics before they become arguments, educational agents with guardrails that actually hold, and why you should never build this on an app someone else controls.

Born out of a YouTube Kids incident that happened last weekend.

https://hosanxiv.substack.com/p/parentos-an-idea-i-cant-stop-thinking

Google is developing a Claude cowork competitor. Will be interesting to see how they catch up

https://www.timesofai.com/news/googles-vision-for-an-autonomous-ai-agent/

https://x.com/bcherny/status/2044847848035156457

Tips for Opus 4.7

Enable Auto Mode (Shift-Tab in CLI or dropdown) - No need for --dangerously-skip-permissions anymore. Run long complex tasks (refactoring, research, features) without babysitting or constant prompts — safe auto-approval for most commands.

Use /fewer-permission-prompts skill: Scan session history and add safe repeated commands to your allowlist to reduce interruptions.

Leverage Recaps & Focus Mode (/focus): Quick session summaries when returning + hide intermediate steps to see only final results

Tune Effort Level: Set to xhigh for most work, max for hardest tasks (/effort) for better intelligence vs speed.

woooo 2 prompts for my first experiment and already hit 9% of the 5 hour limit 😂😂😂

EDIT: testing on my claude pro sub-account (15/mth trial for 3 months)

https://x.com/claudeai/status/2045156267690213649

Claude Design: Figma Killer?

for those of you who are into genAI, here’s a useful prompt gallery site

https://veosoragen.com/

Codex users, try this out

gives you more aesthetic front-end websites instead of the usual AI website templates.

https://github.com/Leonxlnx/taste-skill

https://x.com/ohryansbelt/status/2045873788051415287

if you are using vercel, pls rotate your .env keys.

sunday night - becoming tech support 😭😭😭

TL;DR: Vercel Security Incident report by the CEO

• A hacker got into Vercel after an employee’s work account was compromised through another AI tool (context.ai)

• The attacker was highly sophisticated, significantly accelerated by AI accessed some internal systems but customer data impact was very limited.

• Vercel is now tightening security, adding new tools, and telling everyone to update their secret keys.

Feels like someone got early access to Mythos.. I don’t think we are ready for what’s to come

https://x.com/rauchg/status/2045995362499076169?s=46