Kuro

All things AI, tech and finance. Looking for logic in lines of code.

[Kuro @0xKurro]

the best vibe coding demo i've seen this year isn't an app.

it's a TV.

the input: a model number.
the output: a TV that went from a sponsored ad before Netflix to something that looks like an Apple TV.

in between, Claude Code:

→ found the TV on wifi
→ removed 25 Xiaomi packages
→ wrote a new launcher from scratch

7M views because everyone has this TV and everyone hates that ad.

backup first, your own device only, network debugging off when you're done.

[Gathin (Ø,G) @gathint]

Vibe coded my parents' TV this afternoon!

It was showing a sponsored ad before you could even get to Netflix.

Gave Claude Code the model number, it found the TV on wifi, removed 25 Xiaomi packages and wrote a new launcher.

Looks like an Apple TV now lol

[Kuro @0xKurro]

16 seconds. watch an app get taken apart and rebuilt without touching its original code.

that's the whole idea behind REA:

the original → a spec → your own version

it's open-source, it plugs into your agent in one command, and it runs the same five stages you see in the dashboard.

investigate → decompile → spec → build → verify

full course in the article i'm quoting. start with one feature, not the whole app.

study the original. ship your own.

(only on software you're allowed to analyse. animation and numbers are mine)

[codila @0xCodila]

REA is most dangerous thing on the internet right now...

It lets your AI agent reverse engineer any app, down to the binary

If you use it correctly, you can rebuild any feature you pay for, for FREE

In this article, I show you how: https://t.co/3P1FNXiXzl

[Kuro @0xKurro]

Opus 5.5 + REA is bad news for every app

REA opens any app on your computer and shows how every feature really works, with the exact place in the code behind each answer

3 steps and you are in:

1. Install REA inside Claude Code with one command.

2. Pick an app you use and one feature you want to understand.

3. Send this to Claude: "Use REA to find out how this feature really works. Show me the proof, then build my own version and test it."

Stop paying for expensive tools when you can literally see what's happening under the hood and build your own alternative.

[codila @0xCodila]

REA is most dangerous thing on the internet right now...

It lets your AI agent reverse engineer any app, down to the binary

If you use it correctly, you can rebuild any feature you pay for, for FREE

In this article, I show you how: https://t.co/3P1FNXiXzl

[Kuro @0xKurro]

🤖 How to train AI agents to review code without getting stuck in "infinite review"?

The team at Anthropic recently shared brilliant insights on how they handle code review using Claude Code. The most fascinating challenge they faced? Preventing AI agents from endlessly reviewing and correcting each other.

📉 The Reality Check: They discovered that without strict boundaries, one single Pull Request (PR) went through more than 50 rounds of reviews overnight. The agents just kept passing the code back and forth.

To fix this loop, they introduced a hard rule: after 3 rounds of review, if no consensus is reached, the system stops and asks the human author to reconsider the overall architectural approach.

🛠 The 4 pillars of the Claude Code review process:

• Pre-PR Verification: Code validation and security checks start before the developer even opens the PR.

• Double-Check Layer: Other AI agents validate findings to significantly reduce false positives.

• Risk-Based Human Review: Human intervention is dynamic—it scales strictly based on the risk level of each specific change.

• Continuous Feedback Loop: Any post-review incidents or edge cases are automatically converted into evaluations (evals) to continuously improve the AI reviewer.

[Kuro @0xKurro]

@undefinedKi So devs at Anthropic are basically just babysitting AI agents now? 👀

[Kuro @0xKurro]

@Bober_smart Obsidian + Claude Code is an absolute cheat code for managing information 💥