@MengTo — building a full video editor with Claude AI
I'm building my dream video editor with Claude AI
11k lines of code, 685 commits, 3 full weeks. Cursor, Tailwind, React, Firebase. Web-based, cloud saves, versions, can edit on mobile. This video was recorded using the tool.
Never in a million years I thought I was capable but
11k lines, 685 commits, three weeks — those are real numbers, not vibes. The stack (Cursor + Claude Sonnet + React + Firebase) is the exact combo everyone's been talking about this summer, and this is one of the cleaner examples of someone actually shipping something non-trivial with it. I'm skeptical of "I built X in a weekend" posts as a genre, but 685 commits over three weeks is a genuine engineering effort, not a weekend hack.
@iannuttall — git alias for committing constantly while using Cursor
Quick Cursor tip for shipping fast: add this to your zshrc/bashrc and commit changes OFTEN
alias save="git add . && git commit -m 'x' && git push"
Sometimes it takes you down a rabbit hole it's hard to get out of 😰
Blunt advice, and it's correct. When you're iterating with an AI assistant, the context window is your working memory and git is your undo history. Committing aggressively before a big refactor or a risky prompt is exactly the right discipline. The alias is a bit rough (git commit -m 'x' will make your history look like a crime scene), but the underlying principle holds.
@RayFernando1337 — live coding session with Cursor + Claude Sonnet 3.5
Building app with Cursor, Claude Sonnet 3.5 code stream
A livestream of someone actually building with Cursor and Claude Sonnet 3.5. Less about the content itself and more about the workflow becoming something people watch in real time. The AI-assisted coding stream is a genre now.
@supabase — Python officially supported
Python is now officially supported
https://supabase.com/docs/reference/python/start
The Supabase Python client going official matters more than the tweet lets on. Before this, the Python SDK was community-maintained and lagged behind the JS client in features and reliability. If you're building data pipelines or ML workflows that touch Supabase, this removes a real friction point. Worth checking the docs to see what's now stable.
@tom_doerr — Sherlock at 53k GitHub stars
Sherlock is pretty popular at over 53k GitHub stars though I wonder what exactly the use case is
sherlock is a Python tool that hunts usernames across hundreds of social networks from a single CLI command. Tom's reaction is fair — "53k stars for a username hunter" is a slightly strange cultural artifact. Most of the stars are probably from OSINT researchers, recruiters, and people who want to check if their handle is available everywhere before launching a product. At 53k it's also clearly used for less savory purposes, which is probably why Tom's wondering. The codebase itself is clean and worth a look if you've never seen how these site-list-based scanners work.
@rubenhassid — Google's prompt library, top 5 picks
Google put together a prompt library.
I've tested all of them.
Here's my top 5 prompts you can try ↓
#1. Blog Post Creator
Honestly saved this to check if Google's prompt library has anything more structured than the usual "act as an expert in…" stuff. The fact that someone actually tested all of them is the useful part — filtering a large prompt library down to five concrete picks is the work.