🌍 Dev.to

Can You Build an Alternative to LLMs? 8 Months, ~200 Failed Experiments, One Wall. 2

This is part 2 of a research series documenting an attempt to build something adjacent to — and in some ways alternative to — large language models. Part 1: "My Synthetic Eval Said 30/30. LoCoMo Said 0.13." The code that survived lives in AuraSDK . Not a chatbot wrapper. Not another prompt stack. Not a vector database with a nicer UI. The question was narrower and more dangerous: Can a non-neural, CPU-only system accumulate experience, mutate its own internal state, and imp...
🌍 Dev.to

Why AI agents fail reasoning tasks: Token Clustering Theory

This article was originally published on BuildZn . Everyone's hyped about GPT-4o and Opus. Amazing for chat, sure. But when my AI agents fail reasoning tasks on the daily, the hype feels like hot air. I've spent weeks debugging weird logical breakdowns in multi-step AI flows, and it’s not just "hallucinations." Something else is going on. Why AI Agents Fail Reasoning Tasks: My Gut Feeling I’ve shipped 20+ apps, built FarahGPT (5,100+ users), a complex AI gold trading syste...
🌍 Dev.to

The Mall of React: An Uncle-Nephew Guide

One mall. Many shops. From the day everything had to be rearranged by hand, to the smart, modern mall you shop in today. Setting the Scene 👦 Nephew: Uncle, I use React every day, but if someone asks me how it actually works inside, I go blank. 👨‍🦳 Uncle: Then let's fix that with one picture in your head: a big shopping mall. Every React idea maps to something in this mall. Keep this table nearby. In the Mall In React The mall building The web pa...
🌍 Dev.to

Two signatures are better than one — bilateral provenance for AI agents

Two signatures are better than one — bilateral provenance for AI agents An AI agent produces a financial report. You notarize it — a 239-byte cryptographic record, signed by an independent notary, anchored on NEAR. The record proves: this hash existed at this timestamp, and $0.01 USDC was paid for the attestation. A week later, someone asks the obvious question: who proved the agent wrote the report? Nobody. The notary signed a hash the client submitted. Anyone could have submitted...
🌍 Dev.to

Extracting Speech Segments with Silero VAD and ONNX Runtime

Hello, everyone. Have you ever wanted to keep only the parts of a recording where someone is speaking? Finding silence before transcription can reduce downstream work and divide a long recording into more manageable pieces. Today, I will use the ONNX model from Silero VAD to detect speech in a roughly 14-second conversation and extract each segment as a WAV file. What I Tested This lab uses FFmpeg to decode an MP3 conversation between two speakers into a 16 kHz mono wavef...
🌍 Dev.to

A Cautionary Rusty Tale

Hi! My name is Sergio Tortosa, and I made my AI-powered documentation generation ...in Rust. Take into account I started this a year and a half ago, when LLMs weren't as advanced as they are today. So I hope this will serve as a story to learn both some technical info as well as what to do (or rather, what not to do) when starting a startup; this should also serve as a kind of evolution of the viability of Rust for web projects. Was this a good idea? You'll find out. About the projec...
Warp goes open source, betting agents and community can outpace closed rivals
🌍 Dev.to

Warp goes open source, betting agents and community can outpace closed rivals

Warp , the developer tooling startup behind the modern terminal of the same name, is open-sourcing its client — and tying that move to a broader shift in how it believes software will be built. With AI agents now capable of handling much of the implementation work, Warp argues the real constraint now lies in defining what to build, coordinating tasks, and verifying outputs. Opening up the codebase, it says, allows a wider pool of contributors to take on that role — effectively supervising a g...
HackerNews

The Log Is the Agent

Article URL: https://arxiv.org/abs/2605.21997 Comments URL: https://news.ycombinator.com/item?id=48790912 Points: 29 # Comments: 2
HackerNews

My ASN Journey series (2024)

Article URL: https://www.animmouse.com/p/my-asn-journey/ Comments URL: https://news.ycombinator.com/item?id=48791256 Points: 15 # Comments: 2
HackerNews

Moby Dick Workout

Article URL: https://www.hogbaysoftware.com/posts/moby-dick-workout/ Comments URL: https://news.ycombinator.com/item?id=48791266 Points: 25 # Comments: 7
HackerNews

Binary Coverage the Wrong Way

Article URL: https://redvice.org/2026/coverage-the-wrong-way/ Comments URL: https://news.ycombinator.com/item?id=48791308 Points: 5 # Comments: 1
HackerNews

Megawatts by Microwave

Article URL: https://computer.rip/2026-07-04-microwave-and-power.html Comments URL: https://news.ycombinator.com/item?id=48791591 Points: 13 # Comments: 0
HackerNews

EV Batteries Are Defying Expectations After Miles

Article URL: https://www.wsj.com/business/autos/ev-batteries-are-defying-expectations-after-hundreds-of-thousands-of-miles-8579de13 Comments URL: https://news.ycombinator.com/item?id=48791755 Points: 19 # Comments: 10
🌍 Dev.to

Nobody is monitoring Bluesky, so I built a mentions scraper for it

I wanted to know when people mention a brand on Bluesky. Simple ask. Turns out Brandwatch, Mention, Hootsuite, basically every social listening tool, still doesn't cover it. They're all busy with X and Instagram while Bluesky sits at 27M+ monthly users. So I looked at doing it myself and found out something most people miss: you don't need to scrape anything. Bluesky runs on the AT Protocol, which is open by design. Public posts are searchable through a documented endpoint. No login, no API k...
🌍 Dev.to

I Opened 3 Security Issues on Microsoft AutoGen and LlamaIndex. Here Is Why

I just opened 3 security issues on two of the most popular AI agent frameworks on GitHub (combined 110K+ stars). The Issues microsoft/autogen#7917 : Docker code executor mounts host filesystem into sandboxed containers without trust boundary validation — container escape vector. microsoft/autogen#7918 : Agent self-modification patterns in Canvas memory module — agents can alter their own operating constraints during execution. run-llama/llama_index#22245 : 441 inst...
🌍 Dev.to

Checking whether ChatGPT actually recommends your product

Ask ChatGPT or Perplexity "what's the best note-taking app" and you get a shortlist of three to five names. Either you're on it or you don't exist in that channel. And buying research keeps moving there. People call measuring this GEO or AEO tracking now. The way most teams do it is pasting questions into chatbots by hand and eyeballing the answers. That stops scaling at about ten questions, and you can't trend it week over week. Doing it programmatically Don't scrape the chat...
🌍 Dev.to

Osloq — ให้ AI reproduction เวลาเกิด bug

Osloq — ใช้ AI หาสาเหตุ bug แทน เวลา AI coding tools เสนอจะ "fix bug ให้" — เราได้แต่กด Accept หรือไม่ก็ Reject สองปุ่ม สองทางเลือก แต่เราไม่เคยรู้ว่า: AI รู้ได้ยังไงว่า bug เกิดจากตรงนี้? มัน reproduce แล้วหรือแค่อ่านโค้ดแล้วเดา? ถ้าเรา accept — มันจะพังของอย่างอื่นไหม? Osloq เลือกทางที่สาม: ไม่ใช่ "fix ให้" — แต่ " หาให้เจอแล้วบอกว่าเกิดอะไรขึ้น " Osloq คืออะไร Osloq เป็น AI agent ที่ทำหน้าที่ "นักสืบ bug" มีคนเปิด GitHub Issue → Osloq อ่าน → trace ...
🌍 Dev.to

Exporting any Bluesky profile's followers with the open API

Every big social network locks audience data behind auth walls and anti-bot systems. Bluesky went the other way. The AT Protocol is open by design, so public profile data (bios, follower counts, full follower and following lists) is queryable through a documented API without logging in. The whole surface is basically two endpoints: GET https://api.bsky.app/xrpc/app.bsky.actor.getProfile?actor=HANDLE GET https://api.bsky.app/xrpc/app.bsky.graph.getFollowers?actor=HANDLE&limit=100 ...
🌍 Dev.to

AgentGuard vs Semgrep vs CodeQL: 100 Percent vs 0 Percent on AI Agent Security

I ran the same 39 AI agent security samples through three scanners: AgentGuard, Semgrep, and CodeQL. The Results Scanner Detection Rate False Positives AgentGuard v0.6.4 100% (39/39) 0 Semgrep 0% (0/39) 0 CodeQL 0% (0/39) 0 Zero. Semgrep and CodeQL detected nothing. They have zero rules for AI agent security. AgentGuard has 17 detection rules covering all 10 OWASP ASI categories plus 4 novel attack vectors: Memory Po...
🌍 Dev.to

Add a post-quantum readiness gate to your CI in 5 lines

Your codebase almost certainly relies on RSA and elliptic-curve cryptography — TLS, JWTs, SSH keys, signed tokens. All of it is breakable by a large enough quantum computer (Shor's algorithm), and "harvest now, decrypt later" means data you encrypt today can be captured today and decrypted later. Regulators noticed: CNSA 2.0 (US federal + suppliers), DORA (EU financial entities, applies from Jan 2025), and NIS2 now mandate strict cryptographic risk management — which in practice means kn...