Insights
Engineering perspectives on architecture, product development, AI, and the everyday decisions behind useful software.
From the engineering desk
Browse practical articles or follow the latest technology updates.
Subscribe via RSS →Curated links from external sources — not 360Softy original articles.
A customer asks your support agent whether their refund went through. The agent checks, says yes, and cites a confirmation number. The refund actually bounced back twenty minutes ago, but the lookup the agent ran hit a store that only syncs overnight....
Comments
We've all done it: pasted an API key into a file "just to test," then a week later after the secret is already committed. So I built Secret Guardian, a VS Code extension that catches secrets live, in , the moment they appear — and visually masks them so they never show 👉 Secret Guardian on the VS Code Marketplace (free) Detects 17+ secret types as you type: AWS, GitHub, GitLab, Google, Slack, Stripe, OpenAI, SendGrid, Twilio, npm tokens, private keys, JWTs, credentials in URLs. A generic high-e
The idea Prop firm traders (people who trade with other people's money) need specific calculators that account for strict drawdown rules. Every existing calculator is either ugly, full of ads, or doesn't know that FTMO has a 5% daily drawdown limit. So I built PropWise — 10 interactive tools across 168 SEO-optimized pages, all in vanilla HTML, CSS, and JS. No framework. No build step. No dependencies. 10 core tools: Position Size Calculator (pre-configured for 12 prop firms) Challenge Cost Cal
I use Claude Code every single day. At my day job, it's my main agent, and I've talked plenty about how much I love it. But on my side projects, I like to play around with other agents: Codex, OpenCode, whatever new CLI shows up that week. And recently I hit a wall that turned out to be more annoying than I expected. I wanted to run a bunch of agents in parallel. Not one at a time, but several, across different projects, at the same time. And the moment I tried to do that, I realized every tool
I've built enough projects to know that environment variable validation isn't glamorous — until your app silently uses undefined as a database URL. There are a bunch of tools for this now, and picking the wrong one means either fighting your tool or fighting your own config. Here's how they stack up. Tool Size Dependencies CLI Framework support CtroEnv under 5 KB gzip 0 Yes Node, Vite, Next.js Zod + manual 50 KB+ 0 (Zod itself is ~13 KB gzip, plus your glue) No You build it envalid 8
Public pushback against data centers has emphasized their water and energy consumption, and now Nvidia is highlighting its claim that the Rubin generation reference design for a fully liquid-cooled data center has "eliminated massive amounts of power usage and pretty much all water usage." Still, it doesn't address all of the concerns around AI data centers, including during their construction, and for the power generation requirements of the massive facilities. Also, as Gizmodo points out, Nvid
Originally published on hexisteme notes, part of a series on building and running an AI agent fleet. To stop an AI research or RAG agent from presenting its own inferences as retrieved facts, split the work so the LLM never decides what is a fact: let the LLM only extract and summarize, and let a deterministic, non-LLM pipeline do all scoring, cross-checking, and labeling. Tag a claim FACT only when a rule is satisfied — corroboration by ≥2 independent sources, or one official API — and downgrad
Let’s start with a conversation
An idea, a challenge, or a system that needs to work better. We’ll help you understand the next step.
Prefer email? [email protected]
