DEV.to · 精选 RSS

Prelude IBM Bob v2.1.0 has a great feature that allows you to "natively" read and edit Office files, such as docx , xlsx , and pptx . I decided to use a .docx file as a "spec" (specification) to see if this is an effective way to prompt or instruct an AI agent. I believe many of us have heard about the "Spec-First" approach—driving AI development with a list of specifications and breaking down coding tasks from them. Context I have an existing project that is broken or unfinished due to limited

A nightly job of mine reads a day of application logs, asks a model to tag each line as noise or signal, and writes the keepers to a summary file. It ran on a small always-on box, called an HTTP endpoint, and printed done into a log I only open when something feels wrong. On the second morning of a 48-hour experiment, the summary file was zero bytes and the log said done — twice. So which one is lying: the job, the shell, or me? That question cost me two days and three wrong theories, and the an

You pushed your .env file to GitHub. Maybe you noticed a minute later, maybe a week later. Either way, you open the repo, see your database password sitting in plain text, and your stomach drops. Your first instinct is to delete the file and push again. Don't stop there. Deleting the file doesn't remove it from your Git history, so the secret is still in an old commit. Anyone who cloned or forked the repo in the meantime still has it. The good news is that the fix is a short list of steps, and t

The cheapest place to control a model-drafted documentation change is the prompt boundary, not the pull request. Before any request leaves your machine, split the page into sections a human must own, sections a model may draft, and sections neither should touch. Then gate that split with a script small enough to read in one sitting and fast enough to run in CI. What follows is a two-command workflow built on a single-file Python script with no third-party dependencies. It produces a manifest des

Most Halloween decoration scripts assume you can edit your <body> tag. On site builders you usually can't, and a script that does document.body.classList.add(...) in the <head> throws, because the body doesn't exist yet. I ran into this while writing the install guides for halloween.js , a small library that draws spider webs in the corners of the page, blinking eyes, a flying witch, a dropping spider and a tombstone. It is MIT licensed, has no dependencies and weighs about 9 KB gzipped. The lib

Japan's Statistics Bureau (Ministry of Internal Affairs and Communications) and the National Statistics Center (NSTAC) publish an extraordinary open dataset called SSDSE (System of Social and Demographic Statistics) . Specifically, SSDSE-A covers all 1,741 municipalities across Japan, aggregating 128 comprehensive public indicators including demographics, households, economic activities, labor forces, municipal finance, healthcare facilities, and education. However, if you simply pass this datas

A step-by-step path from your first tdk up to running a full multi-service stack on your machine. Work through the stages in order. Each stage builds on the one before it, and you can check off topics as you go. Learning path for TDK CLI . Practice projects are in the TDK example repositories . flowchart TD A[Start: Prerequisites] --> B[Stage 1: First Stack] B --> C[Stage 2: service.json] C --> D[Stage 3: Generated Files] D --> E[Stage 4: Running and Inspecting] E --> F[Stage 5: Multi-Service St

I was a frontend engineer at Sybill from May 2024 to September 2026. Founders do not wait for a ticket. For two and a half years, neither did I. In May 2024 I joined Sybill, a Silicon Valley startup that builds AI for sales teams. I was hired as a frontend engineer. Over the next two and a half years I built a large part of Ask Sybill, our AI chat, from its first version to the day I left. A strong backend team built the AI brain behind it. My job was everything around that brain, so that people

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle. What I Built It is 10:30 PM. A prospective enterprise client emails you a 14-page RFP under a strict mutual Non-Disclosure Agreement (NDA). They need a fixed-price proposal and a milestone plan by 9:00 AM tomorrow. As a freelancer or boutique agency engineer, you are caught in a bad dilemma: The Dangerous Shortcut: Paste the client's confidential brief into a commercial cloud chatbot (OpenAI, Anthropic). You

By Rokas Remeika. Published 2 October 2026. When multiple tenants share an inbound phone number, the caller's phone number is the only routing key available. If a caller exists in multiple accounts, a database query that takes the first match will route the call non-deterministically based on index order. The fix is to gather every matching record across all tenants assigned to the dialed number, then sort them using an explicit four-level tie-break chain ending in the unique record ID. Writing

Disclosure: I build dochost, the service used in the examples. The setup steps are specific to it; the four rules in the second half apply to any "publish" tool you give an agent. Checked against the live server on October 18, 2026. Most of what an AI assistant writes for other people ends up as a file: a Markdown report, an HTML quiz, a status page. Then someone has to put it somewhere with a URL. If the assistant can call a publish tool, that step goes away. You ask for the page, the assistant

I often use AI assistants when working with websites and web applications. Sometimes I need to ask a question about a page I'm looking at, but the AI has no idea what's actually on my screen. I can copy the text and paste it into the chat, but that doesn't always work well. On a simple article page, it's usually enough. On a dashboard or a settings page, things get more complicated. A lot of useful information is lost: which elements are buttons, which are links, whether a menu is open, or how d

The Intro Idempotency. Eight letters, five syllables, and a word I pretended to understand in meetings for an embarrassingly long time. It kept showing up in API docs, in Stripe's dashboard, in that one coworker's code review comments. I nodded. I did not know. So today we are going to build the bug that makes idempotency matter, charge a fake customer twice, and then fix it. By the end, "idempotent" will be a word you can use in a sentence without sweating. The What Here is the dictionary versi

In 2012, Knight Capital lost $460 million in just 45 minutes due to an incomplete software deployment and dead code execution. Reusing an old feature flag across eight servers allowed an unverified legacy routine to execute 4 million unauthorized trades. Strict deployment verification and aggressive dead code deletion are essential safeguards for modern production systems. When I study classic software engineering disasters, the 2012 Knight Capital collapse stands out as the ultimate cautionary

Most guests book their stay from a phone. They often scroll on mobile data, sometimes from an airport gate, and they make up their mind in a few seconds. A hospitality website gets judged by different rules than a portfolio or a company page: it either answers "what will I get, and how do I book it?" right away, or the visitor goes back to a booking platform. I am a WordPress and Flutter developer in Famagusta, North Cyprus. Many of the small hotels, villas and holiday rentals around me depend o

I used to tell my coding agent something like this: Update develop/1.65.0 . Create feature/<task key> from it. Implement the task. Then make an MR into develop/1.65.0 . Tomorrow the same release could already be in alpha , and a bit later in beta . So I had to know the current stage of every release myself and pass the right branches to the agent every time. The release stage was the one piece of my release process that lived outside Git, CI and Jira: in people's heads and in a Confluence table

Claude Code can add a task to a team board and show Ana's open tasks as a table. It doesn't have a plugin for that. It talks to a Java server, built with Spring AI, that exposes the board over the Model Context Protocol (MCP). This article builds that server step by step: first as a handful of annotated methods, then locked down with OAuth 2, a permission for every tool, and validation on every argument. Getting it to work takes minutes. Making it safe is most of the job. There's an animated vid

By Jesalyn Sual | 1st-Year BS Computer Science Student Introduction Hello, everyone! I am Jesalyn Sual, a first-year Bachelor of Science in Computer Science (BSCS) student at PLMun. As a Computer Science student, I am beginning my journey into the world of programming and technology. One of the programming languages I am learning is C++. At first, programming seemed difficult because I needed to understand new terms, symbols, and rules. However, as I continued practicing, I realized that program

Originally published at tensorust.com . TL;DR: On October 17, 2026 I put my open-source Rust LLM proxy, tokio-prompt-orchestrator , between the official OpenAI SDK and a fake provider, and measured what it actually does. 200 identical requests fired at once became 1 upstream call. During a 60 second provider outage, the circuit breaker cut the calls that hit the dead API from 89 to 5 . And the same test found a bug I had shipped: the server sends one upstream call at a time , so 20 different pro

In Dashy's Hindi translation, someone translated a variable name. The message said {थीम} instead of {theme} . The JSON was valid and the diff looked fine. But vue-i18n could not compile the message. In Hindi, clicking Save in the theme editor stored the colors and then threw an error, so no confirmation appeared and the editor stayed open. I found the same kind of bug in two other projects: Open WebUI: placeholder names translated in 10 locales. listmonk: five Arabic messages lost their {name} .

Deterministic formatting normalization prevents expensive LLM generation retries when a model output fails structural validation due to minor whitespace variations. Quick solution: function normalizeQuickSolutionLabel ( content : string ): string { return content . replace ( /^ ([\*\_]{2} Quick solution: [\*\_]{2})\s * ( .+ ) $/gm , ' $1 \n $2 ' ); } When a publishing pipeline enforces strict layout rules, a valid model answer often fails review solely because a required label shares a line with

A buyer pays 100 Stars, Telegram takes them, and the bot replies to the payment message with "Sorry, I don't know this command". The on_paid handler that should hand over the goods never runs, and there is no row for the payment anywhere. That is what my library, aiogram-stars, did the day before its 0.1.0 release, in any bot that included a catch-all router before it. That bug never made it into a release. Before I published the first version, I went through it again with one question: where ca

Udhaar Khata: The Notebook Behind the Counter What I Built In almost every kirana store in India, there is a thin notebook next to the cash box. Page after page, it records entries like Ramesh ₹500, Sita ₹300, and Amit ₹1,000 baaki. This is the udhaar khata , the informal credit ledger that small businesses use to track customer debts and repayments. Tutors maintain similar records for pending fees, while tailors, milkmen, chemists, and neighbourhood shops rely on trust and handwritten notes to

Udhaar Khata: The Notebook Behind the Counter What I Built In almost every kirana store in India, there is a thin notebook next to the cash box. Page after page, it records entries like Ramesh ₹500, Sita ₹300, and Amit ₹1,000 baaki. This is the udhaar khata , the informal credit ledger that small businesses use to track customer debts and repayments. Tutors maintain similar records for pending fees, while tailors, milkmen, chemists, and neighbourhood shops rely on trust and handwritten notes to

Helpdesk: "Patch Tuesday was Tuesday. It's Friday, and the WSUS console still says 47 updates are needing files. The Windows 11 cumulative update is approved for the Pilot group, and the pilot PCs sit at 'Downloading - 0%'. Last night's synchronization shows Failed, and the Synchronizations page doesn't say why. A forum thread says to stop WSUS, delete the WsusContent folder, run wsusutil reset, and reinstall the role if that doesn't work. Can we do that before tonight's change window?" Not yet.

AI tools are becoming increasingly capable, but there's still a gap between what AI can do and what most people can actually use. A developer might connect an API to a database, build a workflow with several services, and automate a task that previously took hours. An experienced AI user might configure an MCP server or install an agent skill to give an AI assistant access to additional capabilities. But what about someone who simply wants to summarize a YouTube video, extract insights from a po

Nobody signs in to play a quiz at PubTrivia. Players scan the code on their table , a web page opens, they type a name, and they are in. No account, no app, no email. A hundred strangers in a pub, zero credentials between them. Which means the name they type is doing three unrelated jobs at once, and I did not notice that until the third one bit. Job one: validation, and there is almost none export function validatePlayerName ( name : string , ): { valid : true } | { valid : false ; error : ' in

CogniPrep has two guide clusters built on one metadata type: the interview guides under /interview/<slug> and the assessment centre guides under /assessment-centre-practice/<slug> . Twelve pages, one GuideMeta , and four of its fields are names for the same page. export interface GuideMeta { slug : string ; /** Short label for breadcrumbs and sibling pills. */ label : string ; /** Page <h1>, also used as the hub card title. */ h1 : string ; /** * The <title>, before the root layout appends ' | C

PubTrivia keeps every quiz you have run : the final leaderboard in full, and a question-by-question record of what each table answered, for the landlord who gets a complaint a week later. The data for that is all in the database. The screen that reads it back had a bug I only found by reading two files in the same sitting, and it is the kind that no test failure and no error report will ever hand you. What a session actually records When a host starts a session, the questions are drawn by Postgr

Nearly every practice card on CogniPrep's public pages shows a still of the thing in play. The SHL hub has one per test, and the numerical reasoning format page reuses the same images across providers. They are not mockups and they are not hand-built. They are browser screenshots of the real screens. Which raises a problem. The real screens need a signed-in account with the right entitlements, they are reached through a dashboard, and they are the last place you want a capture script authenticat

From Service Account to Root: What CVE-2026-102490 Actually Buys an Attacker Vulnerability overview Much of the Zammad discussion focuses on the first stage. CVE-2026-102489 provides session hijacking and remote code execution as the zammad user, needing no privileges and only some passive user interaction. CVE-2026-102490 is the step that matters afterwards: local privilege escalation from that service account to root. Individually scored 8.5 on CVSS 4.0, the pair is rated 9.4, Critical, and bo

When coding agents such as Claude Code and Codex became good enough to take whole tasks, the first reaction in our industry was very loud: developers would be replaced, and soon. A year later the picture is more interesting than that. Agents write a large share of the code now, and that is not going back. But it is also becoming clear that someone still has to know what correct code looks like, and that this someone is a developer with a critical mindset. In my view we are close to the point whe

When an HTML5 video requests fullscreen playback inside an Android WebView, the application often encounters visual artifacts, broken back button handling, or frozen system bars upon exit. Quick solution: Implement WebChromeClient callbacks to store the original layout visibility and window flags before adding the custom video view to a dedicated FrameLayout container, and restore those exact flags inside onHideCustomView. private View mCustomView ; private WebChromeClient . CustomViewCallback m

The five-second video existed, but the embedded UI had not caught up. Its completion message and receipt field did not update, and its exit button remained disabled. This happened in a private prototype running Wan2.2 TI2V-5B on an L4 in Google Colab. The useful result was more specific than “the tool worked”: generation, receipt, and runtime release each had separate evidence. Inspect the existing job before submitting another The trial generated one silent clip of a paper boat moving across wa

Hey DEV Community! 👋 I'm Shamil — A Developer Who Loves Building Things Hello everyone! 👋 I'm Shamil TP, a developer from Kerala, India, and a first-year MCA student at St. Joseph's College (Autonomous), Devagiri. I enjoy exploring technology, building projects, solving problems, and figuring out how things work behind the scenes. For me, programming isn't just about writing code — it's about turning ideas into something real. And this is my first post on DEV! 🚀 💻 What I Work With I'm interested

If you have ever tried to answer "what do people complain about in our category?" by reading Reddit, you know the useful part is rarely the post. It is the comment thread underneath, three or four levels deep, where someone explains why they returned the product and five people reply with what they bought instead. This post walks through how to get those threads into an AI agent as structured data, using MCP for chat clients and a plain REST API for scripts. I work on ProLevelMax, the data sourc

HIVE Robots is testing humanoid robots for overnight shelf replenishment in a Føtex supermarket in Denmark. The Denmark-based startup is combining Unitree-based humanoid hardware with virtual reality teleoperation, allowing remote pilots to guide robots through product placement and replenishment at real store shelf heights. The live trial is significant because it shows a practical route from human-guided robot operation to retail autonomy, but the robots are not yet independently working in cu

Building RecallAI in 4 Days: What We Tried, What Went Wrong, and What We Learned Have you ever explained something to an AI chatbot, returned to it a few days later, and had to explain everything all over again? That was one of the problems we wanted to explore with RecallAI: an AI tutor that can remember useful information from previous conversations and use it to give more personalized responses. I'm Aman Kumar, and along with my friends Rishu,Abhishek and Ayan, I worked on RecallAI for Graph

Encuentra el cuello de botella, haz que la verificación sea repetible y elige herramientas de IA que ayuden a que los cambios útiles lleguen a los clientes. En los primeros días de la IA, cuando todavía no teníamos estas herramientas como Cursor, Claude Code o Codex, recuerdo que copiaba y pegaba código en una interfaz de chat web y obtenía los snippets de código para usarlos en mi editor. Por aquel entonces no sabía cuánto podíamos confiar en la IA o no. Así que recuerdo que esto me hacía más r

Every form on PubTrivia's sign-in page leads somewhere I have had to be careful about since: Continue with Google, email and password, and the "Forgot password?" link underneath them. This is about the third one, and about one line that was awaited and then thrown away. Here is the line, as it shipped: await supabase . auth . exchangeCodeForSession ( passwordData . code ) const { error } = await supabase . auth . updateUser ({ password : passwordData . password }) if ( error ) return { message :

Notifio is a desktop app that watches rental search pages and emails you the moment a new listing appears. It ships as two macOS DMGs and one Windows installer from the download page , built with electron-builder out of a 140-line package.json . I went through that file line by line last week, not looking for anything in particular, and found four names in it that refer to nothing. Every one of them has been there for months. Every build since has succeeded. That is the interesting part. Not tha

Notifio is an Electron app. The main process runs a monitor over rental search pages, a local HTTP server, and a reply engine. The renderer is a React app that talks to that server over 127.0.0.1 and an SSE stream. Nothing else. It has two files called types.ts . src/types.ts has 26 exported types in 350 lines, and src/renderer/types.ts has 16 in 164. Ten names appear in both files. That is duplication, it is deliberate, and I spent an afternoon last week convincing myself it should stay. This i

I build DevCleaner, a Mac app for clearing developer caches: Xcode DerivedData, simulators, Docker, node_modules and similar. The roadmap is public and voting decides order. Five new ideas: 1. Multiple versions view. Most Macs I look at have three Xcode-era simulator runtimes, two JDKs and several Node versions nobody uses. One list grouped by toolchain, with a suggestion which version your projects still need. 2. Startup items manager. Launch agents and daemons with the app that owns them, a to

Since May 2025 the U.S. Department of Energy has used Federal Power Act section 202(c) to keep eight power plants running past their planned retirements, with orders that last about 90 days and are renewed as they expire. EIA's monthly generator inventory, Form EIA-860M, asks every plant owner for a planned retirement month. I compared the two, month by month, from May 2025 to August 2026. The question was simple: for the units under these orders, is the "planned retirement month" in EIA-860M an

You build a login endpoint. It checks the password, and now you need to remember that this particular visitor is logged in the next time their browser shows up — which, on a server, is a completely new, unrelated request. Reach for useState and it resets every time, because it only lives for one server render. Reach for a module-scope variable and you've just reintroduced the cross-request leak this series already covered. So you either stand up a database table for sessions, or reach for a thir

A QA ticket landed with a screenshot: "Submit button missing on checkout." I opened the page in Chrome. The button was right there — blue, rounded, impossible to miss. I opened it in Firefox. Same button. I assumed a bad screenshot and closed the ticket. Three days later it came back, same screenshot, same page, from a different tester. This time I asked what they'd changed. "Nothing," they said. "I just use Windows with High Contrast on, because I'm light-sensitive." I flipped that setting on m

Memory Systems End Goal is to create AI agent . Agent is like a orchestrator + Integration of LLM + Action of Another tool + Memory. We need to give MEMORY to the Agent . Memory means --> இதுக்கு முன்னாடி நான் இது கிட்ட என்னென்ன கேட்டு இருக்கேன் & அது கொடுத்த விடை. [ Query & Response/Answer ],--> This combination is going to sit in the MEMORY. How to store ? We cant't store continuously. What is the LIMIT ? There are types , lets discuss one by one. BUFFER MEMORY For each chat , we are going to

Let’s talk about something I see over and over in Web3 and AI right now. Everyone wants to build AI agents. And when builders decide to give an AI agent access to data, many reach for the same default stack: Chunk documents, compute embeddings, store them in a vector database, and run cosine similarity search (RAG). For many tasks, like answering questions about documentation or searching blog posts, vector search works just fine. But the moment your problem depends on how things connect, vector

If you ever saved an OpenAI, Anthropic or other provider key in an OpenCode workspace, create a replacement key today, deploy it, then delete the old one and watch usage for 48 hours. On 17 October 2026 the OpenCode team said it "identified unauthorized access to a legacy system that we were in the process of decommissioning" and that third-party provider keys saved in workspaces "may have been exposed"[1]. The Hacker News thread passed 850 points within a day. This post is a rotation checklist,

Every dataset eventually produces a value that makes you stop and ask, "is that real?" A $9,000 order in a column of $40 orders. A heart rate of 300. A test score of 120 out of 100. Before you average, model, or plot anything, you need a defensible way to flag those values. The IQR method (Tukey's fences) is the most widely used rule for this. It's simple, it needs no assumption about the shape of your data, and it's what most box plots are drawing under the hood. This post covers: What the IQR