Sentinel: an agentic fraud investigator on TigerGraph
🌍 Dev.to

Sentinel: an agentic fraud investigator on TigerGraph

Repo: github.com/v4xsh/sentinel Every month, a bank's card-fraud team opens twenty investigations that a human still has to run to the ground. Was this transaction fraud. What kind. How far did it spread. Do we file a SAR. My submission to Hacker House Goa × TigerGraph is Sentinel, an agent that reads six months of the Vesta card-transaction graph and produces, for each flagged alert, a verdict, a fraud-pattern label, the actions the bank should take with §2 approval routes, and a FinCEN-sh...
Picking a New Task's Project in 200 ms: Jev vs Claude on 53 of My Tasks
🌍 Dev.to

Picking a New Task's Project in 200 ms: Jev vs Claude on 53 of My Tasks

Introduction Open Walnut is a self-hosted web app for running Claude Code sessions. Every session sits on a task, and the task board shows which sessions are running and which are waiting for you. When you type a new task, Walnut tries to fill in its project, priority and pinned tier while you are still typing. Since September 2026 those guesses can come from Jev, a decision-only model from TypeSafe AI, instead of a Claude model. On 53 of my own tasks with a known project, Jev picked th...
A GUI JavaScript Operating System For The ESP32
🌍 Dev.to

A GUI JavaScript Operating System For The ESP32

Haris16-code / KryonOS KryonOS is an ESP32 GUI operating system with a JavaScript runtime for apps. It provides a real touch-based interface and allows installing and running JS apps dynamically without reflashing firmware. KryonOS KryonOS is an open-source , lightweight, high-performance GUI Operating System and JavaScript App Runtime designed specifically for the ESP32 microcontroller. It provides a complete desktop-like...
🌍 Dev.to

Decentralized Code Reviews in Git: Introducing git-notes v0.1.8

Code review is one of the most critical steps in software development, yet it remains heavily locked into centralized web platforms. When you rebase a branch or work offline, discussions are scattered or lost. We built git-notes to bring code reviews back into Git itself. 💡 How It Works git-notes leverages Git's built-in refs/notes/* reference architecture: Comments : refs/notes/comments Reviews & Approvals : refs/notes/review Todos : refs/notes/t...
🌍 Dev.to

Redact the Review Packet Before a Model Sees an OSS Failure Log

A small open-source patch often dies in review for reasons that never appear in the diff stat. A contributor copies a failing CI log into a chat box and asks a model to propose a fix. The log still holds a short-lived fixture token and an absolute path from the runner home directory. The returned diff also deletes a comment that explained why a locale-sensitive comparison had to stay explicit. The pull request then looks cleaner than the bug, and that kind of cleanliness is a poor form of pro...
🌍 Dev.to

ShellCanvas: an open-source graphical workspace for SSH servers

ShellCanvas is a free, open-source desktop app I’m building for managing remote servers over SSH. It brings a file manager, terminal sessions and a text editor into one graphical workspace, with a separate workspace for each host. GitHub · Downloads and screenshots · Documentation How it works The interface runs on your computer. Connect using your existing SSH account, then open files and terminals in movable windows. You do not need to install a ShellCanvas agent on th...
Choosing Between MethodChannel and PlatformView: A Real Decision, Not a Rule of Thumb
🌍 Dev.to

Choosing Between MethodChannel and PlatformView: A Real Decision, Not a Rule of Thumb

I Embedded a Native Camera Preview in Flutter (And Learned Exactly Where MethodChannel Stops Being Enough) Seven years into mobile engineering, I wanted to answer a question I've seen a few times: when do you actually need PlatformView instead of MethodChannel? Most answers I've seen are either "PlatformView is for native UI" (true, but not useful) or skip straight to copy-pasteable boilerplate with none of the failure modes that actually teach you anything. Over the weekend, I built a sm...
Congrats to the Summer Bug Smash Winners!
🌍 Dev.to

Congrats to the Summer Bug Smash Winners!

We are thrilled to announce the winners of DEV's Big Summer Bug Smash powered by Sentry ! This challenge asked you to find your biggest rockstar bugs and give them an epic grand finale, and you delivered. We got real pull requests against real codebases, performance wins measured and verified, and debugging stories that had us cheering in victory. Thank you to everyone who participated. Now, let's celebrate our winners! 🐞 Clear the Lineup Winners The Bug T...
🌍 Dev.to

Macula Mesh+Realm: Architecting Survivability in a Contested Space

Introduction Macula Mesh+Realm is a sovereign network substrate that provides secure, identity-based connectivity across untrusted networks. Unlike traditional VPNs, which create tunnels to extend existing networks, Macula creates a distributed fabric where identity and routing are cryptographically integrated. The Technical Foundation: Beyond the Tunnel Macula is not built on legacy tunneling protocols. It leverages a modern, high-performance stack to ensure survivability. ...
🌍 Dev.to

RelayPlane vs LiteLLM vs OpenRouter: an honest comparison for cost caps and routing

If you want a hard daily spend cap and a per-request cost ledger on one machine with no database server to run, RelayPlane is the smallest install of the three ( @relayplane/proxy on npm, free, MIT). If you want 100+ providers behind one gateway for a team, LiteLLM is the better tool. If you want a hosted API you don't run yourself, OpenRouter is the better tool. All three are real, none of them is a straight replacement for the other two, and this piece is an honest breakdown of where ...
We Tested a 35B LLM Against Typed-Decision Models on 12,000 Real RFQs—Confidence Changed the Winner
🌍 Dev.to

We Tested a 35B LLM Against Typed-Decision Models on 12,000 Real RFQs—Confidence Changed the Winner

91.9%. 89.6%. 78.0%. Those were the primary-class accuracies of a typed-decision API, a 35B mixture-of-experts LLM, and a 421M open-weight decision model on the same real classification job. But accuracy was not the result that changed the deployment decision. The decisive question was: Can the model tell us when its answer is safe enough to automate—and when a human needs to look? To find out, I compared Jev , Qwen3.5-35B-A3B , and Laya 421M on 12,000 U.S. federal IT sol...
🌍 Dev.to

Cisco Secure Email Gateway Injection Bug CVE-2026-76443: What Administrators Should Do First

Cisco Secure Email Gateway Injection Bug CVE-2026-76443: What Administrators Should Do First Vulnerability overview CVE-2026-76443 affects Cisco Secure Email Gateway and Cisco Secure Email and Web Manager, versions 15.5 and earlier. It is an improper neutralization vulnerability, published as part of a five-CVE hardening release in September 2026. CERT-In rated the release CRITICAL. The defect allows attacker-supplied input to reach security-sensitive processing contexts without...
🌍 Lobsters

F-Droid 2.0

Comments
HackerNews

AI Workers' Inquiry 2026

Article URL: https://techworkersinquiry.org/ai/ Comments URL: https://news.ycombinator.com/item?id=49834235 Points: 2 # Comments: 0
HackerNews

Show HN: bananabread AI – Find what customers love and want from your products

We recently launched an API that finds what people consistently love about a specific consumer product. Some of our first users reached out asking whether it could also find what people still want from the product. That’s what we're launching today. For example, reviews like "I love the taste, but I wish it had less sugar" tells you what's working well already and what you could improve. Give the API a brand + product name (only consumer goods work for now). It first searches for thousands of re...
HackerNews

Show HN: DrawCMS – An open-source animated diagramming tool for AI Agents

Hello, I'm building DrawCMS. It's an open source animated diagram editor you can self-host. You can ask a compatible AI agent to read your repository and create an editable diagram, then walk through the flow step by step on the canvas. AI makes it easy to write code quickly. Keeping track of how that code works can be harder. You finish a feature, move on, and a week later you're trying to remember where a request goes, which service writes to the database, or why a background job runs. It supp...
HackerNews

Don't let AI kill the author

Article URL: https://www.economist.com/leaders/2026/09/24/dont-let-ai-kill-the-author Comments URL: https://news.ycombinator.com/item?id=49834680 Points: 2 # Comments: 0
HackerNews

We turned off AI access for new hires

Article URL: https://www.valon.ai/blog/we-turned-off-ai-access-for-new-hires Comments URL: https://news.ycombinator.com/item?id=49834932 Points: 3 # Comments: 0