Загрузка видео...

Не удалось загрузить видео

На главную

closed source is dead... i just reverse-engineered CCleaner / CleanMyMac with Opus 5.5 for Mac & Windows PC $40/year cleaners → replaced in one afternoon. → free. open code. no account. no upsells. how it came together: • step 1 → Claude Code on Opus 5.5 in an empty...

34,415 просмотров • 1 день назад •via X (Twitter)

Комментарии: 13

Фото профиля Codez
Codez1 день назад

You are my senior desktop engineer and product designer. Together we are shipping "Cleanly", a free, open-source disk cleaner for macOS and Windows that looks and feels like Apple built it. I attached screenshots of CleanMyMac and CCleaner: treat them as a quality bar only. Name, logo, icons, copy and code must be 100% original. Work in phases. At the end of every phase, show me what you made and wait for my "go" before the next one. ─────────────────────────────────── PHASE 1 — Design exploration (no app code yet) ─────────────────────────────────── - Make 3 clearly different directions as clickable HTML mockups of the home screen: 1. light glass, 2. dark neon, 3. minimal / editorial. - Each mockup has: - a sidebar; - a hero card with a giant "X GB ready to reclaim" number and one Clean button; - a disk "space map" of bubbles sized by folder size; - 4 small status cards: Cleanup, Speed, Protection, Apps. - Make them interactive: Clean counts the number down, the bubbles react. - Then iterate with me on fonts, icons and buttons until it feels premium: - display font for numbers, clean UI font, mono font for small labels; - a custom duotone icon set; - a calm glass "capsule" primary button. Deliverable: one approved concept page. ─────────────────────────────────── PHASE 2 — Architecture & contract ─────────────────────────────────── - Stack: Electron (latest) + React + TypeScript + Vite. esbuild bundles the main process (ESM) and the preload (CJS). - Write `shared/types.ts` first: every data type and the full typed API between the main process and the UI. The main process's handler map must be typed so that a missing method fails the build. - Security baseline: - sandbox: true, contextIsolation: true, nodeIntegration: false; - the preload exposes only the typed API; - IPC refuses calls from any frame that is not our own index.html; - only https links leave the app, and they open in the system browser; - strict CSP in production, fonts bundled offline. - A mock implementation of the whole API, so the UI runs in a normal browser with realistic fake data (npm run demo). Deliverable: the contract, an app shell with the sidebar and routing, the design system in CSS, and the browser demo running. ─────────────────────────────────── PHASE 3 — Build in parallel ─────────────────────────────────── Split the work across sub-agents. Each owns its own files, and no two touch the same file: A. File scanner + delete/trash + disk map + large files B. Duplicates, startup apps, installed apps, protection status C. Settings (keys encrypted with Electron safeStorage), AI assistant, tips, feedback D. Home screen E. File screens (Cleanup, Space Map, Large Files, Duplicates) F. System screens (Speed, Apps, Protection, Assistant, Settings) G. Tip + Feedback screens and the mock API H. App icon, installers, CI What each part must do: - Scanner: - async walker using lstat, ~16 concurrent fs operations, never follows symlinks or junctions, bounded memory, progress events to the UI; - categories: - Windows: temp, Windows temp, browser caches for every profile, app caches, crash dumps, update downloads, dev caches, Recycle Bin, downloads untouched for 90+ days; - macOS: Library caches, logs, browser caches, Xcode DerivedData, dev caches, Trash, old downloads. - Space Map: - bubbles sized by folder size, drill-down, a source switch: Home plus every fixed drive (C:, D:, external volumes on Mac); - bubbles are a physics toy: drag, throw, collide (mass by area), spring back. Write transforms straight to the DOM, with no React re-render per frame. - Startup apps: - show impact (high / medium / low); - toggling writes the same StartupApproved REG_BINARY value Task Manager writes; - system-wide items go through one elevated reg.exe behind a UAC prompt; - on macOS, launchctl for LaunchAgents. - Apps: - real icons, sizes, last used; - uninstall with the vendor's own uninstaller (Windows) or move the .app to Trash (macOS). - Protection: - read-only check of Defender / firewall / Gatekeeper / XProtect / SIP with a 0–100 score; - honest copy: "Cleanly does not replace your antivirus". - AI assistant: - Claude via the official Anthropic SDK, or ChatGPT; the user's own key; - send only names, paths and sizes, never contents; - get structured JSON verdicts per path: safe / review / keep. - Tips: a native in-app screen with preset amounts, one-time / monthly and a note. Payment opens Stripe Checkout in the system browser. The app never sees card data. - Feedback: idea / bug / love, star rating, message, optional email and system info. Sends to <YOUR CONTACT EMAIL> by opening the user's mail app with the letter ready. ─────────────────────────────────── PHASE 4 — Safety review (separate agent, adversarial) ─────────────────────────────────── Its only job is to find ways real user data could be lost or leaked, and fix them. It must cover at least: - Never trust paths from the UI. The main process re-derives every cleanup location from the category id. - Canonicalize the FULL path with realpath, then run the guard again. Refuse: - drive roots, the home folder itself, OS and Program Files folders; - $RECYCLE.BIN, System Volume Information; - UNC/network paths, alternate data streams, trailing dots/spaces; - 8.3 short names like DOCUME~1; - standard user folders even when they were moved to D:. - Caches are deleted for good; user files ALWAYS go to the system Trash. - Keep temp files younger than 24 h (running apps use them). - Old downloads are processed after the Recycle Bin is emptied, never before. - File icons: no network/removable drives, and never .url/.lnk/.scf files. Windows would authenticate to a remote host and leak an NTLM hash. - Elevated actions: build arguments only from server-side data, escape for PowerShell, and handle a cancelled UAC prompt. - reg.exe prints in the OEM code page (e.g. 866): decode it properly. Deliverable: a list of every issue found and how it was fixed. ─────────────────────────────────── PHASE 5 — Integration & real-machine test ───────────────────────────────────- typecheck + build fully green. - Click through every screen in the browser demo: no console errors, no unhandled rejections, no effects without cleanup. - Package the Windows app and run it on my PC READ-ONLY. Capture screenshots through the Chrome DevTools protocol (--remote-debugging-port) and show me real numbers: disks, reclaimable GB, startup items, apps. Do not press Clean without my OK. - Known traps, check them before they bite: - backdrop-filter over an animated canvas → 3 fps. Don't. - WebGL shaders: render at 0.3–0.5× resolution and use real elapsed time per frame. - PowerShell 5.1 loses the exit code of an elevated child → re-read the registry instead of trusting it. - NSIS "could not open file" → the path is over 260 chars; build from a short folder. - Single-instance lock: a test run needs its own --user-data-dir. ─────────────────────────────────── PHASE 6 — Ship ─────────────────────────────────── - electron-builder: - Windows: NSIS installer + portable .exe; - macOS: .dmg + .zip for arm64 and x64, ad-hoc signed (hardenedRuntime off until we have an Apple ID). - Installer file names WITHOUT the version (Cleanly-setup.exe, Cleanly-mac-arm64.dmg…), so `releases/latest/download/<name>` links never go stale. - App icon: generate it from an SVG (squircle, soft pastel gradient, our brand mark). - GitHub Actions on a "v*" tag: build Windows on windows-latest and macOS on macos-latest, then publish a GitHub Release with every installer. - Before making the repo public: scan the whole git history for keys, tokens, emails and local usernames. Add a LICENSE (MIT). - README: features, the safety model, how to build, and first-run notes: - Windows: SmartScreen → More info → Run anyway; - Mac: xattr -cr /Applications/Cleanly.app. ─────────────────────────────────── PHASE 7 — Website (in parallel with phase 6) ─────────────────────────────────── - A separate session builds an Apple-style landing page in the same design language: - hero with a live recreation of the app window; - scroll-story sections: scan, space map, safety, AI, speed, protection; - FAQ and a download section with a Contact link. - Download buttons read from one config file that points to releases/latest/download. - Deploy to Vercel. Lighthouse 95+ on mobile and desktop. No horizontal scroll at 390px. ─────────────────────────────────── Definition of done ─────────────────────────────────── ✓ the app runs on my PC and finds real junk on every drive ✓ the safety review is complete, with fixes listed ✓ a public GitHub Release with .exe and .dmg that anyone can download ✓ the website is live and its download buttons work ✓ screenshots of the real app, not the mock

Фото профиля Josh Hoeg | Vibe Coding & AI
Josh Hoeg | Vibe Coding & AI1 день назад

$40 a year to empty a cache folder was always wild pricing

Фото профиля Codez
Codez1 день назад

Yeah. Also true

Фото профиля Beef Supreme
Beef Supreme1 день назад

The moat will be trust

Фото профиля Movez
Movez1 день назад

dammn, this is sick !

Фото профиля Codez
Codez1 день назад

thx mate !

Фото профиля no0x
no0x1 день назад

Claude just killed paid cleaners in one afternoon!

Фото профиля Fahrenheit
Fahrenheit1 день назад

Paid cleaners charge for what's free. The real flex is the agent whose only job was to break it.

Фото профиля BuildWithVenkey
BuildWithVenkey1 день назад

thanks bro. you just nailed it.

Фото профиля Addy
Addy1 день назад

Love this build. Inspired by REA, I re-built a REA skill with recursive self learning, so it keeps building till the task is done. Give it a spin!

Фото профиля ☠️ Ryan ☠️
☠️ Ryan ☠️1 день назад

It's nice to be able to do but the real game changer is doing it in a low level language with only a few external packages. Less bloat. Less resources the app needs. Don't just fight the war on pay walls. Fight the war on bloat.

Фото профиля Android 🔳
Android 🔳1 день назад

Windows so much wasted tokens. Reverse engineer it to Linux instead 😄

Фото профиля kimchipatriot
kimchipatriot1 день назад

I thought Claud stopped this from happening?

Похожие видео

I just reverse-engineered Apple AirDrop with Opus 5.5 to send files between my Mac & Windows PC this is insane... drag a file onto the other computer → hop. it's there. no cloud. no account. no cable. → here's the build step by step: • step 1 → open Claude Code on Opus 5.5 in an empty folder and paste the prompt below • step 2 → Opus picks the stack: one Electron codebase that runs on macOS and Windows • step 3 → discovery: every 1.5 s each computer says "i'm here" over UDP on your local network. the other one pops up on a radar • step 4 → transfer: files stream directly between the two machines. the other side gets Accept / Decline. "always accept" for your own computers • step 5 → before any UI, Opus wrote a test: two fake devices find each other and send a folder + a 3 MB file, byte for byte • step 6 → design: Opus split the work across parallel agents. one rebuilt the iOS 17 NameDrop glow as a WebGL shader, one drew 3 icon concepts and picked the best, one built the download site • step 7 → separate reviewer agents checked every piece. one caught a shader bug that would've rendered black on the Mac. i'm on Windows, i'd never have seen it • step 8 → GitHub Actions builds the .dmg and the .exe on every release. the Mac build runs on Apple hardware in the cloud, not on my PC • step 9 → first launch on the Mac: the icon just bounced in the dock. Opus added a startup log and a fallback that forces the window open. next build opened fine one afternoon. zero lines written by me. drop it on the Mac. it lands on the PC. that's it... find full prompt below ↓ Then read the article on how to reverse-engineer anything with Opus 5.5.

Movez

53,756 просмотров • 3 дней назад

this is crazy... the guy reverse-engineered his Meta Ray-Bans 3 with Opus 5.5 into a 3,000+ Elo chess engine he built AI chess cheat glasses on Opus 5.5 + Jev and open-sourced the whole workflow here's how it works step by step: step 1 → install it on your Mac with one command: npx chesshacker setup. It downloads Stockfish from GitHub and verifies the checksum step 2 → put on Meta Ray-Bans and video-call your own Mac on WhatsApp. The glasses see the board, the Mac sees what the glasses see step 3 → the Mac grabs frames straight from the call window. About 1 per second while it waits for the opponent, 1 every 3 seconds on your turn step 4 → Jev runs a cheap pre-check on every frame: is the board readable, is a hand over it, did a piece move? No change means no expensive AI call step 5 → only then Opus 5.5 reads the board. The app compares it with the position it already knows and accepts a move only if it's legal step 6 → Stockfish thinks for half a second and picks the best reply, with an eval like +1.5 or "mate in 3" step 7 → macOS speaks it through a virtual mic into the WhatsApp call: "Rook a to d1". You hear it in your ear step 8 → on the Mac a command center shows everything live: whose turn it is, the suggested move with a confidence bar, the move list and a log under the hood: → if Opus loses track of the board, the glasses say "Lost track of the board" and you fix it with undo or by typing the move → if Stockfish crashes, it restarts once and says "Something went wrong. Still trying" → every game is saved as a PGN file, API keys live in the macOS Keychain → an overlay draws the suggested move as an arrow on the live video, ready for OBS Clone his workflow below Then read the guide on how to reverse-engineer anything with Opus 5.5.

Movez

392,501 просмотров • 4 дней назад

Jev + Opus 5.5: Anthropic's new model beats GPT-6 Astra for 1/5 the cost, and 4 API changes will 400 your agent before it writes a single line I pulled these 10 steps from the migration docs so you don't learn them in production step 1 → $4 / $20 per 1M. Opus 5 was $5 / $25. cache reads dropped from $0.50 to $0.20 step 2 → 66.4% on Terminal-Bench 4.0 vs GPT-6 Astra 57.9% and Opus 5 52.3%. +14.1 points in one release, and on FrontierCode it beats Astra at default effort for 1/5 the cost step 3 → thinking can't be turned off anymore. send thinking: disabled and you get a 400. drop the field, set effort step 4 → tool_choice any and tool are gone. 400. switch to auto + strict step 5 → edit anything above a thinking block and the request dies. append only, or opt into drop_block step 6 → computer_20251124 is dead on the API. 400. move to computer_toolset_20260801 step 7 → the quiet one: default effort fell from high to medium. your agent thinks less than you set it up to and nothing tells you step 8 → hop Opus 5.5 → Sonnet 5 → Opus 5.5 and you pay 4.36 instead of 3.32. +31%, the cache dies and Sonnet can't read Opus's reasoning step 9 → change effort at the top of the request and the cache is gone. Jev sets it per message and the cache stays step 10 → switch fast - standard mid-session and it's a full cache miss. Jev picks speed once, on turn one one model, three knobs, zero 400s. that is Jev + Opus 5.5 send this to your Claude Code before you touch the model ID, then read my full Jev deep dive in the article below ↓

Carnage

16,674 просмотров • 1 месяц назад

THIS IS A PURE F*CKING GOLDMINE Anthropic just shared one of the best tricks for Claude Code. The idea: let Opus 5.5 write the plan as a script, and let a swarm of cheaper agents run it. Bigger jobs and way fewer Opus tokens, because the script holds the plan and Opus only sees the final answer. You just launch Claude Code like this: claude --model opus --effort ultracode here's the split: → Opus 5.5 turns your task into a workflow script → Haiku 5.5 agents scan every file in parallel → Sonnet 5.5 agents patch each fix in its own isolated copy → you stop burning Opus context on a hundred agent transcripts Opus only shows up at three moments: → writing the workflow before the run → reading the verified results after it → signing off before anything ships Opus plans. Haiku scans. Sonnet ships. JEV engineering is the same move one layer down: the forks that need no thinker (which file, which tool, retry or stop) go to JEV in milliseconds, and the big models only see the ones that split the full tree > Opus 5.5 runs the main session and writes the workflow > scan audits every file on Haiku 5.5 > edit patches each fix on Sonnet 5.5 in an isolated copy > verify cross-checks every finding on Sonnet 5.5 > Opus 5.5 reads only what survived verification > JEV routes micro-forks in milliseconds paste the tree and this prompt into Claude Code ↓ "Rebuild my Claude Code setup around this tree: 1. Run the main session on Opus 5.5 with ultracode on. 2. In every workflow, run read-only scan stages on Haiku 5.5 and edit stages on Sonnet 5.5, each edit in its own isolated copy. 3. Add one rule to CLAUDE.md: every finding gets an independent verifier agent before it's reported, and fix loops stop after two rounds with no progress. 4. Route fast deterministic forks (tool choice, path selection, retries) through JEV. Show me every change as a diff first. No edits until I say go." ↳ save it, it'll come in handy.

xsmryze

60,207 просмотров • 8 дней назад

THIS IS A PURE F*CKING GOLDMINE Anthropic just shared one of the best tricks for Claude Code. The idea: let Opus 5.5 write the plan as a script, and let a swarm of cheaper agents run it. Bigger jobs and way fewer Opus tokens, because the script holds the plan and Opus only sees the final answer. You just launch Claude Code like this: claude --model opus --effort ultracode here's the split: → Opus 5.5 turns your task into a workflow script → Haiku 5.5 agents scan every file in parallel → Sonnet 5.5 agents patch each fix in its own isolated copy → you stop burning Opus context on a hundred agent transcripts Opus only shows up at three moments: → writing the workflow before the run → reading the verified results after it → signing off before anything ships Opus plans. Haiku scans. Sonnet ships. JEV engineering is the same move one layer down: the forks that need no thinker (which file, which tool, retry or stop) go to JEV in milliseconds, and the big models only see the ones that split the full tree > Opus 5.5 runs the main session and writes the workflow > scan audits every file on Haiku 5.5 > edit patches each fix on Sonnet 5.5 in an isolated copy > verify cross-checks every finding on Sonnet 5.5 > Opus 5.5 reads only what survived verification > JEV routes micro-forks in milliseconds paste the tree and this prompt into Claude Code ↓ "Rebuild my Claude Code setup around this tree: 1. Run the main session on Opus 5.5 with ultracode on. 2. In every workflow, run read-only scan stages on Haiku 5.5 and edit stages on Sonnet 5.5, each edit in its own isolated copy. 3. Add one rule to CLAUDE.md: every finding gets an independent verifier agent before it's reported, and fix loops stop after two rounds with no progress. 4. Route fast deterministic forks (tool choice, path selection, retries) through JEV. Show me every change as a diff first. No edits until I say go." ↳ save it, it'll come in handy.

delost

399,685 просмотров • 16 дней назад

this is wild... a guy reverse-engineered his Meta Ray-Bans 3 with Opus 5.5 and turned it into a 3,000+ Elo chess engine it’s AI chess cheat glasses built on Opus 5.5 + Jev, and the whole workflow is open source here’s the full breakdown: step 1 → one command sets it up on your Mac: npx chesshacker setup it pulls Stockfish from GitHub and checks the checksum step 2 → put on the Ray-Bans and start a WhatsApp video call to your own Mac the glasses look at the board, and the Mac sees exactly what they see step 3 → the Mac captures frames straight from the call window about 1 per second while it waits for your opponent, 1 every 3 seconds when it’s your move step 4 → Jev runs a cheap filter on every frame: can it read the board? is a hand in the way? did a piece move? if nothing changed, the expensive AI call is skipped step 5 → only then does Opus 5.5 read the board it checks the result against the position it already has and accepts a move only if it’s legal step 6 → Stockfish thinks for half a second and returns the best reply with an eval like +1.5 or “mate in 3” step 7 → macOS reads the move aloud through a virtual mic into the call: “Rook a to d1”. it plays right in your ear step 8 → a command center on the Mac shows it all live: whose turn it is, the suggested move with a confidence bar, the move history and a log under the hood: → if Opus loses the board, the glasses say “Lost track of the board”. you fix it with undo or by typing the move in manually → if Stockfish crashes, it restarts once and says “Something went wrong. Still trying” → every game is saved as a PGN, and API keys are stored in the macOS Keychain → an overlay draws the suggested move as an arrow on the live video, ready to drop into OBS grab the workflow below 👇 then read the guide on reverse-engineering anything with Opus 5.5

Codez

640,988 просмотров • 3 дней назад