I'm not just a vibe coder. I'm an engineer who uses AI to ship real, useful software.
Ukraine
Joined March 2013
- Tweets2K
- Following106
- Followers78
- Likes8K
Ivan Tugay retweeted
This is about a new TypeScript compiler. I had a modest budget — just two days of a subscription — and my weekends, and I wanted to see what it actually takes to get a comparable result. In the end, on key benchmarks it came out no worse, and in some places better, than teams with significantly larger resources. The difference isn't the money, it's how wisely you use it. So tell me — should I keep going with this experiment, or just leave everything as is? Would anyone be interested in me continuing, or is it better to stop?
Ivan Tugay retweeted
@satyanadella These models are not human. By creating them in our own image, we embed the very same flaws. There is no need to make models more human; they should be tools.
Ivan Tugay retweeted
Manage your local multi-cloud environment using Floci UI, a unified web console for developers. It consolidates AWS, Azure, and GCP emulation into a single interface that renders real infrastructure state rather than mock data.
github.com/floci-io/floci-ui
Ivan Tugay retweeted
App Store Connect CLI 3.6.0 is out!
Main improvement is better API contract handling, so commands preserve Apple's response fields and reject unsupported requests.
Also: fixes for App Clips, Game Center, TestFlight, webhooks, offers, and analytics.
github.com/rorkai/App-Store-…
Ivan Tugay retweeted
🔗 Microsoft Store:
apps.microsoft.com/detail/9p…
💻 GitHub:
github.com/Deepender25/Edge-…
Ivan Tugay retweeted
Your agent's memory is quietly lying to you.
(it paraphrases your words instead of keeping them)
MemPalace holds your conversation history as exact text and retrieves it with semantic search. It doesn't summarize, extract, or paraphrase.
Instead of compressing what you said into tidy notes, it keeps the words and indexes where they live.
Why that matters: summarizing is where memory loses the thing you'll need.
- The detail that gets dropped is the one you go looking for a month later
- You remember the conversation happened; the summary can't rebuild it
- A flat pile of vectors gives you no way to narrow a search
No embedding API, no cloud, nothing leaves your laptop.
mempalace. wire it into Claude Code as an MCP server.
57,933 stars. MIT.
save this before you explain your project from scratch again.
Ivan Tugay retweeted
Brooooooo... This Chrome extension is sooo perfect for LLMs. 🤯
It lets you instantly convert almost any webpage into clean Markdown, making it easier to use with ChatGPT, Claude, Gemini, and other AI tools.
Perfect for AI enthusiasts, developers, designers, researchers, and anyone who works with web content :)
Follow me for more amazing AI, Coding & Web Dev insights 💎
Ivan Tugay retweeted
Just open sourced –
> 𝚗𝚙𝚖 𝚒 𝚠𝚎𝚒𝚐𝚑𝚝𝚕𝚒𝚏𝚝
A lightweight client-side AI model manager for local-first applications.
𝚠𝚎𝚒𝚐𝚑𝚝𝚕𝚒𝚏𝚝 takes model loading, caching and progress reporting off your hands so that you can build browser-based AI experiences faster.
Demo weightlift.dev/
🐙 Github github.com/wassgha/weightlif…
Ivan Tugay retweeted
someone just built a free alternative to cluely. no data stored on their servers.
it's called natively.
same ui as cluely, zero learning curve, runs entirely on your machine.
→ real-time transcription and ai meeting notes
→ undetectable stealth mode
→ local rag and byok
→ works with gpt, claude, gemini, llama, or fully offline via ollama
cluely had a data breach in 2025 that exposed 83,000 users' transcripts, screenshots, and personal info.
natively stores everything locally and has never had a breach.
9,000+ users. $0 for personal use.
Ivan Tugay retweeted
ECC is an operating system for agent workflows that works natively across multiple harnesses and languages.
- Skills, instincts, and memory optimization for continuous learning
- Security scanning and research-first development built in
- Supports 12+ language ecosystems including TypeScript, Python, Go, and Java
- Cross-harness agent workflows from real-world engineering use cases
Explore it here:
osp.fyi/ecc
Ivan Tugay retweeted
My Claude Code setup was one CLAUDE. md file. For months.
And I kept re-explaining the same project conventions in every session, wondering why the output was hit or miss.
Turns out there's a whole project structure under .claude/ that almost nobody uses.
CLAUDE.md is just the entry point, the "project brain" that loads at session start with your stack, architecture, and commands.
• The rules/ folder is where it gets good. Instead of one giant file, you split instructions by topic: code-style .md, testing .md, api-conventions .md. They can even target specific paths in your repo.
• commands/ holds repeatable workflows as slash commands. I run /review and /fix-issue daily.
• skills/ load only when the task needs them, so your main context stays small.
• agents/ are sub-agents with their own isolated context. A code reviewer that doesn't pollute your main conversation.
• hooks/ run before or after tool calls. Mine blocks anything unsafe before it executes.
• And .mcp.json wires up external tools and lives in Git, so the whole team shares the same setup.
I made a visual of the full structure; it's below.
If you set up just one thing from this, make it the rules/ folder.
And if you want to go deeper into agents, skills, and practical AI for .NET developers, join the group here: skool.com/thecodeman-ai-tool…
Ivan Tugay retweeted
every AI coding tool has the same dirty secret: it ships fast, broken code.
you vibe code a feature with Claude/Codex/Copilot, push it, and CI turns red in 5 minutes.
lint errors. missing docs. failed tests. broken commit hygiene.
so you fix. push again. fix. push again.
found a tool that kills this loop entirely: no-mistakes
instead of `git push origin`, you do:
git push no-mistakes
and it spins up a disposable worktree, runs review → test → docs → lint, auto-fixes what's safe, escalates only what needs your judgment — then pushes and opens a clean PR for you.
nothing reaches your remote until every check is green.
works with claude, codex, copilot, opencode, and more — basically a quality gate sitting in front of whatever agent wrote your code.
6k+ stars, MIT licensed, fully open source.
the timing makes sense too — AI is generating more code than ever, but "code that runs" and "code that's actually ready to ship" are two very different things. this is the first tool I've seen actually built for that gap.
link in commen 👇
Ivan Tugay retweeted
⭐️4.7K | #DeepSeek has released a guide for integrating its models into popular AI coding tools
It covers Claude Code, Cline, Codex, GitHub Copilot...
Each guide walks through install, config, and first run — helping developers get DeepSeek-V4-Pro / V4-Flash running in minutes.
Ivan Tugay retweeted
MeshLLM pools GPUs and memory across your home, office, or friends' machines and exposes the whole thing as one OpenAI-compatible API.
> 31 nodes, 5 active models, 277GB of pooled VRAM — no cloud involved
> Models too large for one machine get split across peers automatically
> One endpoint, localhost — the mesh decides where each model actually runs
- No subscription, no cloud bill, just hardware you already own
Your personal GPU cluster was always possible. Someone finally made it a one-node setup.
Github: github.com/Mesh-LLM/mesh-llm
Ivan Tugay retweeted
Your coding assistant phones home more than your browser does.
Free Code is Claude Code with every telemetry ping stripped and 88 experimental flags unlocked. One binary, zero callbacks. No tracking, no reporting, just the tool.
8.5K devs already made the switch.
⭐ 8.6K #OpenSource #Privacy
github.com/freecodexyz/free-…
Follow for daily dev finds 🔔
Ivan Tugay retweeted
Governments pay millions for this.
Someone vibe-coded an intelligence dashboard that tracks military flight paths, satellite fires, radiation levels, sanctions lists, and live conflict events from 27 intel feeds.
And pings your phone the second anything changes.
→ 3D WebGL globe + live world map
→ Two-way Telegram + Discord bots
→ Multi-tier alerts (FLASH / PRIORITY / ROUTINE)
→ Runs 100% locally. No cloud. No telemetry.
here's the wildest part:
hook it to Claude, GPT, Gemini, Grok, or 4 other LLMs and it becomes a two-way analyst. text /brief to it on Telegram from your phone → get a full cross-domain intel readout in seconds. /sweep triggers a live scan.
Bloomberg Terminal is $24,000/year. private OSINT platforms cost more. this is $0.
100% Open Source. 10k stars.
Ivan Tugay retweeted
Nexterm is an open-source server management tool that centralizes remote access via SSH, VNC, and RDP. It facilitates SFTP file management, Docker application deployments, and Proxmox container administration, featuring organizational structures for secure access control.
github.com/gnmyt/Nexterm
Ivan Tugay retweeted
fully uncensored Cybersecurity-specialized model tuned on exploits, pentesting and Run locally on MacBook, delivers expert-level offensive & defensive insights.
- Zero refusals
- Exploit writing
- vuln analysis
- Fully uncensored
- OWASP, and MITRE.
- fine-tuned deep on real security data
- 11 quants available from tiny Q2_K to full precision, Runs great on consumer hardware.
Runs 100% offline on your laptop no refusals.
Perfect for CTFs, bug bounties, blue/red teaming.
