标签：#dev

Most on-page audits catch the obvious stuff: a missing title here, a duplicate meta description there. The thing that quietly costs you rich results is structured data that exists but is invalid, and most flat-list crawlers either skip it or bury it. Here is why it happens and how to catch it. The problem, concretely You add FAQ schema to a product page to win that expandable rich result in Google. You paste a JSON-LD block into the head, ship it, and move on. Six weeks later the rich result never showed up, and nobody knows why. The usual culprits are small and silent: A @type that does not match the content (FAQPage with no mainEntity ). A required property missing ( acceptedAnswer without text ). A trailing comma or a stray character that makes the JSON parse fail entirely. Schema that contradicts what is actually on the page, which Google can flag as spammy and ignore. None of these throw a visible error. The page renders fine. The schema is just dead weight, and a standard "issues" crawl that only counts titles and headings walks right past it. How to catch it First, validate the JSON itself. A block that does not parse is invisible to search engines. Even a quick local check surfaces the dumb-but-fatal errors: // Pull every JSON-LD block and check it parses + has a @type const blocks = [... document . querySelectorAll ( ' script[type="application/ld+json"] ' )]; blocks . forEach (( b , i ) => { try { const data = JSON . parse ( b . textContent ); if ( ! data [ " @type " ]) console . warn ( `Block ${ i } : missing @type` ); } catch ( e ) { console . error ( `Block ${ i } : invalid JSON ->` , e . message ); } }); If that logs an error, the schema was never going to work, no matter how perfect the markup looked. Second, check required properties for the specific type you are using. FAQPage needs mainEntity with Question items, each carrying an acceptedAnswer . Article needs headline , author , and datePublished . Validating "it parsed" is not the same as "it is c

AI coding tools are incredible. But I noticed something — they ship code fast, skip the rules, and nobody catches it until it's already in production. That's exactly what GuardDuo is built to fix. The Problem We're in the age of vibe-coding. You describe what you want, the AI builds it, it works — and you ship it. But "works" and "correct" are two very different things. Imagine asking an AI to build a login form. It works perfectly. But under the hood it has hardcoded API keys, no input validation, missing aria-labels , and it's using fetch directly instead of your project's apiClient wrapper. Your Issue said none of that was allowed. Nobody caught it. That's the vibe-coding trap — and it's happening on every team using AI-assisted development right now. What is GuardDuo GuardDuo is a GitLab Duo Agent skill that acts as your AI guardian. Instead of just reviewing code in isolation, it cross-references your code changes against the actual intent of the linked GitLab Issue — using the Orbit Knowledge Graph , which is essentially the brain that knows your project's rules, requirements, and success criteria. In plain terms: GuardDuo reads what the Issue asked for , reads what the code actually does , and tells you exactly where they don't match. It audits across three dimensions: 🔐 Security — hardcoded secrets, SQL injection, missing input validation ♿ Accessibility — missing alt text, aria-labels , poor color contrast 📐 Standards — deviations from your project's established patterns and conventions And when it finds a problem, it doesn't just flag it — it fixes it. How It Works Just open GitLab Duo Chat or GitLab Agent Platform(on your choice of IDE) -> choose the agent as GuardDuo and type: Audit issue #[issue no.] — GuardDuo pulls the Issue context from Orbit, analyzes the code, and returns a structured report Fix issue #[issue no.] — GuardDuo generates a corrected implementation that satisfies all requirements Or paste any code snippet directly and ask it to audit o

devtools #infrastructureascode #cdk #aws Index TL;DR Major Features Bedrock AgentCore — From Alpha to Stable Fn::GetStackOutput & Weak Cross-Stack References Validations Framework Performance Improvements CloudWatch PromQL Alarms CLI Improvements New L2 Constructs Service Enhancements Community Highlights Community Content & Resources How Can You Be Involved Hey CDK community! Here's an update covering everything that shipped in April and May 2026. TL;DR Bedrock AgentCore graduated to stable — production-ready AI agent infrastructure with semver guarantees. Cross-region references got a major upgrade with native Fn::GetStackOutput support and weak cross-stack references. The new Validations framework replaces policyValidationBeta1 with a richer plugin system. And file fingerprinting is ~33% faster with persistent asset caching. These features are available in aws-cdk-lib v2.247.0 through v2.257.0 and aws-cdk CLI v2.1116.0 through v2.1125.0. Full changelogs on GitHub Releases ( Library | CLI ). Major Features Bedrock AgentCore — From Alpha to Stable The @aws-cdk/aws-bedrock-agentcore-alpha module has graduated to aws-cdk-lib/aws-bedrockagentcore — stable APIs, semver guarantees, production-ready. If you've been building AI agents with Bedrock but held off on CDK because of the alpha label, it's time to upgrade. ( #37876 ) AgentCore provides the core infrastructure for building AI agents: runtimes, gateways, identity management, observability, and online evaluation. The Policy submodule remains in alpha as it continues to evolve rapidly. ┌─────────────────────────────────────────────────────┐ │ Bedrock AgentCore (Stable) │ ├─────────────────────────────────────────────────────┤ │ │ │ ┌──────────┐ ┌──────────┐ ┌──────────────────┐ │ │ │ Runtime │ │ Gateway │ │ Identity │ │ │ │ (L2) │ │ (L2) │ │ (L2) │ │ │ └────┬─────┘ └────┬─────┘ └────────┬─────────┘ │ │ │ │ │ │ │ ▼ ▼ ▼ │ │ ┌──────────┐ ┌──────────┐ ┌──────────────────┐ │ │ │Observa- │ │Online │ │ Policy Engine │ │ │

Why API Breaking Changes Still Reach Production Even With CI/CD A few years ago I watched a "tiny" API change take down checkout for about forty minutes. The change was a one-liner. The pull request had two approvals. CI was green across the board. And it still broke production, because the thing that actually mattered was never tested. If you run microservices at any real scale, you have lived some version of this. Let's talk about why it keeps happening even with a mature pipeline, and what the teams who don't keep getting paged do differently. The Problem Here's the change that caused the outage. A payments service had a response that looked like this: { "status" : "ok" , "transaction_id" : "txn_8842" , "amount_cents" : 4200 } Someone renamed amount_cents to amount and switched it to a decimal, because "cents is confusing." Cleaner field, better docs. The producing service's tests were updated to match, everything passed, it shipped. The problem: three downstream services still read amount_cents . One of them was the order service, which now received undefined , multiplied it by a quantity, and wrote NaN into the database. The failures didn't even surface in the payments service. They surfaced two hops away, in a service the original author had never opened. This is the core issue. A breaking change is not defined by the service that makes it. It's defined by the consumers who depend on it. And the producer's CI pipeline has no idea those consumers exist. Why Existing Approaches Fail The natural reaction is "we need more tests." But look at what each layer actually checks. Unit tests verify the code does what the author intended. The author intended to rename the field. The unit tests were updated to expect amount . They passed because they were testing the new, broken behavior. Green unit tests told us nothing. Integration tests verify the service works with its own dependencies — its database, its cache, the APIs it calls. They almost never spin up the services

Rethinking fake content in Symfony projects A prototype web page displaying pure placeholder content When building early UI prototypes or shaping design systems in Symfony, placeholder content becomes a constant companion. Lorem ipsum text. Dummy profile photos. Placeholder videos. Silent audio. Temporary avatars. Realistic fake user data. Every project needs them — and yet most setups rely on a patchwork of libraries, links and hardcoded values. Omnia Ipsum aims to fix that by giving Symfony developers a single, elegant toolkit for placeholder content of all kinds. In this article, I will walk you through the motivation behind the project, the conceptual patterns it follows, and its most advanced features — all designed to make your prototyping workflow faster, cleaner and more maintainable. Motivation: Why a placeholder library? Most Symfony projects start the same way: You add lorem ipsum text manually into Twig templates. You grab placeholder images from an external service. You generate avatars using yet another site. You paste in temporary YouTube or stock video URLs. You install Faker separately whenever realistic data is needed. The result is inconsistent, fragmented and difficult to maintain. And even worse: placeholder content often leaks into production unless guarded carefully. The idea behind Omnia Ipsum was simple: “If your UI needs placeholder content, it should come from one place — predictable, configurable, and accessible directly from Twig.” This cuts down on boilerplate, cognitive overhead, and the "temporary chaos" of early-stage templates. Quick Start Prerequisite Go to github.com/symfinity/recipes and follow the instructions to add the required recipe repository. Installation composer require --dev symfinity/omnia-ipsum Usage Use the Twig functions immediately: <img src= " {{ omnia_image ( 600 , 400 ) }} " alt= "Placeholder" > <img src= " {{ omnia_avatar ( 'John Doe' , 100 ) }} " alt= "Avatar" > <video src= " {{ omnia_video ( 1920 , 1080 ) }}

The Problem Typography should be one of the simplest parts of a project. In reality, it often ends up scattered across multiple layers: Bootstrap: $font-family-base variables Tailwind: JavaScript configuration TypeScript: type definitions Design systems: W3C Design Tokens The same font information gets copied and maintained in several places. Every update means touching multiple files, hoping everything stays in sync. It's repetitive, error-prone, and easy to get wrong. So I built Font Manager. Define your fonts once and export them in whatever format your project needs — CSS, Bootstrap variables, Tailwind configuration, TypeScript definitions, design tokens, and more. The Solution A simple Twig function: {{ font_manager ( 'Ubuntu' , '400 700' ) }} Configuration: symfinity_font_manager : export : formats : - scss_bootstrap - tailwind_config - typescript_definitions One lock command: php bin/console fonts:lock Every format, automatically generated. Perfectly synced. Bootstrap Example Before: // Manually copy font name $font-family-base : 'Ubuntu' , sans-serif ; // ❌ Duplication @import 'bootstrap/scss/bootstrap' ; After: symfinity_font_manager : export : formats : [ scss_bootstrap ] php bin/console fonts:lock // app.scss @import './assets/styles/fonts-bootstrap' ; // ← Auto-generated @import 'bootstrap/scss/bootstrap' ; Bootstrap uses your fonts automatically. No manual mapping. No duplication. Tailwind Example symfinity_font_manager : export : formats : [ tailwind_config ] // tailwind.config.js const fonts = require ( ' ./assets/fonts-tailwind.config.js ' ); // ← Auto-generated module . exports = { theme : { extend : { fontFamily : fonts . fontFamily } } }; <p class= "font-sans" > Your custom font, via Tailwind. </p> TypeScript Example symfinity_font_manager : export : formats : [ typescript_definitions ] import { fonts , type FontFamily } from ' ./assets/fonts ' ; applyFont ( element , ' sans ' ); // ✓ Valid applyFont ( element , ' invalid ' ); // ✗ TypeScript erro

As web developers, engineering workflows are heavily dependent on the NPM registry. However, the traditional process of searching, auditing, and integrating packages remains highly fragmented. Developers are routinely forced to hop between npmjs.com, GitHub source repositories, and external documentation tabs simply to verify bundle sizes, check dependency trees, or generate setup boilerplate. Following a strong reception on LinkedIn, X, and Facebook, the Motion Mind Team has introduced LibPilot to the dev.to community. LibPilot is not a traditional registry interface; it is an AI-powered search engine and discovery hub engineered to index, track, and analyze over 4,000,000 NPM packages in real time. Here is an architectural breakdown of how LibPilot restructures package exploration for modern developers and autonomous AI code agents. Intent-Based Discovery and Global Search Architecture Traditional search engines require users to input the exact name or strict keyword of a library. LibPilot introduces a dual-input architecture on its home page to eliminate this constraint: Direct Registry Querying: Users can input full or partial package names into the global search bar to instantly surface clean, structured, and typed suggestions directly from the live NPM ecosystem database. Contextual AI Recommendations: For scenarios where the ideal package is unknown, developers can type out a complete description of their project architecture or system constraints (for example: "a lightweight, typed state management engine that handles server-side rendering natively"). LibPilot's internal AI agent processes the functional requirements and suggests production-ready libraries suited for that stack. Continuous Context AI and Interactive Onboarding A core goal of the platform is reducing developer friction and maintaining deployment momentum. LibPilot transitions static package documentation into an interactive environment: Unlimited AI Chat Architecture: Once a library is select

Semana passada eu passei três horas debugando um bug que deveria levar 20 minutos. O problema? Um módulo de validação escrito em 2019 que ninguém mexe "porque funciona". Spoiler: não funcionava mais, e quando finalmente abri o arquivo, encontrei um // TODO: refactor this datado de 2020. Por que legacy vira bola de neve A indústria trata código legado como se fosse dívida técnica opcional — algo que você paga "quando tiver tempo". Mas código legado se comporta mais como mofo: se espalha, contamina áreas adjacentes, e quanto mais você ignora, mais cara fica a limpeza. O ciclo é previsível: você herda um projeto ou feature antiga, vê que está "meio bagunçado mas roda", adiciona sua feature com um if a mais, e segue em frente. Seis meses depois, outra pessoa faz o mesmo. Um ano depois, aquele arquivo tem 800 linhas, cinco níveis de if aninhados, e zero testes. Ninguém mais entende o fluxo completo, então cada mudança vira uma sessão de especulação: "se eu mexer aqui, quebra ali?" O custo real de esperar Esse código "que funciona" tem um custo oculto que aparece em três formas: Velocidade de desenvolvimento despenca. Features que deveriam levar dois dias levam uma semana porque você passa mais tempo entendendo o contexto do que escrevendo código novo. Bugs aumentam exponencialmente. Código sem testes e com lógica embolada é um gerador de regressões. Você corrige um edge case e quebra outro que nem sabia que existava. Onboarding vira tortura. Novo dev no time? Boa sorte explicando por que aquele service tem três formas diferentes de fazer autenticação, ou por que a mesma validação está copiada em sete lugares. Sinais de que você está sentado em cima de uma bomba Nem todo código antigo é legacy tóxico. Aqui estão os red flags que indicam que você precisa agir agora: // Red flag #1: comentários mentirosos ou inúteis function processPayment ( order ) { // Process the payment const user = order . user ; // TODO: fix this later // HACK: don't touch this, breaks prod if ( user

If you've ever tried to pull live options data into a Python script, you've probably hit the same wall I did: the cheapest real-time providers start at $99/mo. Here's how to do it for $20/mo — or free if you stay within 1,000 credits/day. What You'll Need Python 3.8+ requests library ( pip install requests ) An API key from market-option.com (free tier available, no card required) Fetching a Full Options Chain import os import requests API_KEY = os . environ [ " MARKET_OPTIONS_KEY " ] BASE_URL = " https://market-option.com/api/v1 " def get_chain ( ticker : str ) -> list [ dict ]: res = requests . get ( f " { BASE_URL } /options/chain/ { ticker } " , params = { " apiKey " : API_KEY }, ) res . raise_for_status () return res . json ()[ " results " ] contracts = get_chain ( " SPY " ) print ( f " { len ( contracts ) } contracts returned " ) print ( contracts [ 0 ]) Each contract in results looks like this: { "details" : { "contract_type" : "call" , "strike_price" : 530 , "expiration_date" : "2026-01-17" , "ticker" : "O:SPY260117C00530000" }, "last_quote" : { "bid" : 3.45 , "ask" : 3.50 , "midpoint" : 3.475 }, "greeks" : { "delta" : 0.42 , "gamma" : 0.031 , "theta" : -0.18 , "vega" : 0.29 }, "implied_volatility" : 0.182 , "open_interest" : 12418 } Filtering by Expiration and Strike def get_near_the_money ( ticker : str , expiration : str , spot : float , width : float = 0.05 ): """ Return contracts within ±width% of spot price. """ contracts = get_chain ( ticker ) low = spot * ( 1 - width ) high = spot * ( 1 + width ) return [ c for c in contracts if c [ " details " ][ " expiration_date " ] == expiration and low <= c [ " details " ][ " strike_price " ] <= high ] atm = get_near_the_money ( " SPY " , " 2026-01-17 " , spot = 530 ) for c in atm : print ( c [ " details " ][ " strike_price " ], c [ " details " ][ " contract_type " ], c [ " last_quote " ][ " bid " ], c [ " greeks " ][ " delta " ], ) Scanning for High IV Contracts def high_iv_scan ( ticker : str , iv_threshold :

If you've shipped anything to the browser, you've used localStorage . And if you've used it for more than five minutes, you've also written this exact line more times than you'd like to admit: const user = JSON . parse ( localStorage . getItem ( ' user ' ) || ' null ' ) The Web Storage API has aged remarkably well for something so small, but it carries three persistent pain points that every frontend codebase ends up papering over by hand. Pain point #1: everything is a string. localStorage.setItem('count', 0) doesn't store the number 0 — it stores the string "0" . Read it back and typeof is "string" . Booleans become "true" / "false" , Date objects collapse into ISO strings (if you're lucky) or "[object Object]" (if you're not), and undefined becomes the literal string "undefined" . So every project grows a thin serialization layer of JSON.parse / JSON.stringify wrappers, plus a pile of defensive try/catch blocks for the day a malformed value sneaks in. Pain point #2: the API is verbose and stringly-typed. getItem , setItem , removeItem — three method calls and a string key for what is conceptually just reading and writing a property. It reads nothing like the rest of your code. Pain point #3: reactivity is broken in the tab you actually care about. The native storage event only fires in other tabs of the same origin. The tab that performed the write never hears about it. So if you want to react to your own storage changes — the overwhelmingly common case — the platform gives you nothing. Stokado is a small, zero-dependency library that addresses all three by wrapping any storage object in a Proxy . It's framework-agnostic, TypeScript-friendly, and works equally well with localStorage , sessionStorage , cookies, async backends like localForage, and a handful of mini-program runtimes. This article walks through what it actually does, feature by feature, with runnable code. Quick start npm install stokado import { createProxyStorage } from ' stokado ' const storage =

You have an API response. You want a Zod schema. So you paste the JSON into a JSON-to-Zod converter, copy the output, and ship it. Here's the trap: a lot of those converters infer basic types only . Your email , your uuid , your url , your ISO timestamp — they all come out as z.string() . The schema compiles, the types look right, and your validator quietly accepts "not-an-email" , "ftp://nope" , and "2026-99-99" forever. I wanted to see exactly how much gets lost, so I ran the same payload through two tools and diffed the output. Everything below is real, copy-pasteable output — nothing edited. The input A pretty ordinary user object: { "id" : "3f2a9c1e-5b7d-4e8a-9f1c-2d3e4f5a6b7c" , "email" : "ada@example.com" , "website" : "https://ada.dev" , "age" : 34 , "rating" : 4.7 , "created_at" : "2026-03-04T10:15:30Z" , "is_active" : true , "address" : { "city" : "Lyon" , "zip" : "69001" }, "tags" : [ "early-adopter" , "beta" ] } Tool 1 — json-to-zod (npm) const user = z . object ({ id : z . string (), email : z . string (), website : z . string (), age : z . number (), rating : z . number (), created_at : z . string (), is_active : z . boolean (), address : z . object ({ city : z . string (), zip : z . string () }), tags : z . array ( z . string ()), }); Structurally correct. But every meaningful field is a bare z.string() / z.number() . This schema will happily validate email: "lol" and created_at: "yesterday" . Tool 2 — TypeMorph import { z } from " zod " ; export const userAddressSchema = z . object ({ city : z . string (), zip : z . string (). regex ( /^ [ A-Z0-9 ][ A-Z0-9 \s\-]{1,8}[ A-Z0-9 ] $/i ), }); export type UserAddress = z . infer < typeof userAddressSchema > ; export const userSchema = z . object ({ id : z . uuid (), email : z . email (), website : z . url (), age : z . number (). int (). min ( 0 ). max ( 150 ), rating : z . number (). min ( 0 ). max ( 5 ), created_at : z . iso . datetime (), is_active : z . boolean (), address : userAddressSchema , tags :

AI has made it easier to produce working software. That part is real. It can write code, draft documents, research a topic, scaffold a prototype, and debug a problem faster than most teams can finish writing a decent ticket. But faster building doesn't automatically mean better product decisions. That's the part I keep coming back to. For decades, software teams optimized around delivery. Requirements, design, development, QA, release. Waterfall softened into Agile. Agile grew into DevOps. The practices changed, but the assumption underneath stayed pretty stable: building software is expensive, so plan carefully before you start. That made sense because, for a long time, it was true. Now that assumption is breaking. AI is doing to software what calculators did to accounting. It isn't eliminating the job. It's moving the job up a level. The syntax, boilerplate, first draft, and some of the debugging are getting offloaded. The work doesn't disappear. The bottleneck moves. Learning is still expensive Here's what didn't get cheaper: understanding what people actually need getting stakeholders aligned deciding what evidence would change your mind putting something real in front of users reading the signal without fooling yourself The old question was: Can we build it fast enough? The new question is: Do we understand the problem well enough? That sounds like a small shift, but it changes the work. It changes what strong engineers spend time on. It changes what product people need from engineering. It changes how teams should define "done." If the code ships but nobody learns anything, did the team actually move forward? Sometimes yes. Often no. Users don't know until they can touch it People are not great at specifying requirements up front. Not because they're difficult. Because they're human. Most of us don't know how we feel about something until we can react to a version of it. A mockup. A prototype. A rough slice. A real workflow with sharp edges. So the fastest pat

Most TypeScript tutorials teach you the language. This article teaches you how to use it. There's a difference. The language has hundreds of features. A real project uses maybe twenty of them regularly, and about eight of them make up the difference between TypeScript that fights you and TypeScript that helps you. These are those eight. Each one comes from a pattern I've seen repeatedly in real codebases: first as an antipattern, then as a realization, then as a habit. The goal isn't to show off advanced type gymnastics. It's to show you the specific things that make your code safer, more readable, and less painful to maintain. TL;DR Most TypeScript pain comes from fighting the type system instead of working with it, any , manual casting, and loose types are the usual culprits. A small set of features, discriminated unions, utility types, satisfies , as const , generics, solve the majority of real-world typing problems. The best TypeScript isn't the most complex. It's the most precise. Table of Contents Tip 1: Use Discriminated Unions Instead of Optional Fields Tip 2: Stop Writing Types Twice with Utility Types Tip 3: Use satisfies to Validate Without Losing Inference Tip 4: Use as const for Literal Types That Don't Drift Tip 5: Write Type Guards Instead of Casting Tip 6: Use Generics to Write Functions Once Tip 7: Use ReturnType and Parameters to Stay in Sync Tip 8: Use unknown Instead of any for External Data Honorable Mentions Final Thoughts Tip 1: Use Discriminated Unions Instead of Optional Fields This is the tip that changes how you model data in TypeScript. Once you see it, you'll spot the antipattern everywhere. The antipattern // ❌ A type that tries to represent multiple states with optional fields interface ApiResponse { data ?: User error ?: string isLoading : boolean } The problem: this type allows impossible states. Nothing stops you from having both data and error set at the same time, or neither set, or isLoading: false with no data and no error . The

TL;DR: An API gateway manages HTTP traffic between services — auth, routing, rate limiting, load balancing for REST and gRPC. An AI gateway manages LLM workloads — token-based rate limiting, model routing, cost attribution, semantic caching, guardrails. Use an API gateway for your microservices. Use an AI gateway for your LLM traffic. Most production teams eventually need both, sitting at different layers. This post walks through exactly where each one fits. When we started adding LLM features to our platform, we already had Kong running for our microservices. The instinct was natural: route the LLM traffic through Kong too. Same auth, same rate limiting, same observability stack. One gateway to rule them all. It worked — for about six months, and only in the sense that requests got through. What it didn't give us was anything useful for actually managing AI workloads. We had no idea what each team was spending on tokens. We had no way to set a budget cap that would fire before the bill arrived. Our rate limits were based on requests per minute, which meant a single request with a 50k token prompt counted the same as one with a 200 token prompt. And when OpenAI had a partial outage, Kong had no concept of "try Anthropic instead" — we just served errors. None of that is a criticism of Kong. It's doing exactly what it was designed to do. The problem was us expecting an API gateway to handle a fundamentally different category of infrastructure problem. Here's the precise distinction, and why it matters architecturally. What an API gateway actually does An API gateway is a reverse proxy that sits between client applications and backend services. It handles the cross-cutting concerns of service-to-service HTTP communication: authentication, authorization, rate limiting, load balancing, SSL termination, request transformation, and routing based on URL paths or headers. A typical request flow through an API gateway: Client sends a request to the gateway endpoint Gateway ve

Why I Stopped Picking AI Models by Hype and Started Picking by Speed Three months ago I almost lost a $14,000 retainer because my chatbot felt sluggish. The client didn't say "your TTFT is too high." They said "it feels dumb." That's freelancer code for "users are bouncing and I'm about to find someone else." I rebuilt that bot in a weekend using a model I'd never even heard of six weeks earlier, dropped average response time from 1.4 seconds to under 300ms, and the client renewed for another six months. That single pivot paid for my rent. So I went down a rabbit hole. I ran the same speed test on every model I could get my hands on through Global API's unified endpoint. Fifteen models. Same prompt. Same regions. Ten iterations each. I'm writing this up because if you're billing by the hour or running a side hustle on a shoestring, speed isn't a vanity metric — it's a profit metric. Let me show you what I found. The Setup (How I Actually Ran the Tests) I'm not a researcher with a rack of GPUs. I'm a guy with a M2 MacBook, a $19/mo Hetzner box, and a stopwatch in the form of Python's time.perf_counter() . Here's how I kept it honest. Date window: All tests run on May 20, 2026 Regions tested: US East (Ohio) and Asia (Singapore) Prompt used: "Explain recursion in 200 words" — boring on purpose, because boring prompts are where most apps actually live Output length: Roughly 150 tokens per run Iterations: 10 runs per model per region, average recorded Streaming: Yes, SSE throughout Endpoint: Global API at https://global-apis.com/v1 I measured two things: TTFT (time to first token — the lag before the user sees anything move) and sustained tokens per second (how fast the words actually arrive after that). Both matter. TTFT is the "is this thing broken?" feeling. Tokens per second is the "is this thing fast?" feeling. Here's the script I used, stripped down to the essentials: import time import requests from statistics import mean API_KEY = " your-global-api-key " BASE_URL

Every quarterly earnings number for every US public company going back to 2009 is sitting in a free, well-documented JSON API run by the US government. No API key. No rate limit for normal use. No paywall. Almost nobody in the dev community seems to know it exists. It's at data.sec.gov , and it's the same data Bloomberg charges $24k/year for. What's in it The SEC requires all US-listed companies to file financial reports in XBRL — a structured XML format where every number is tagged with a standardised concept name. The EDGAR system has been collecting these since around 2009. The companyfacts endpoint exposes all of it as clean JSON: GET https://data.sec.gov/api/xbrl/companyfacts/CIK{cik}.json Where CIK is the company's SEC identifier (10 digits, zero-padded). For Apple, that's 0000320193 . The response is a large JSON object with every concept the company has ever reported, broken down by period. The other endpoint you need is the ticker-to-CIK map: GET https://www.sec.gov/files/company_tickers.json This gives you a flat list of all US-listed companies with their CIK, ticker, and name. Load it once and cache it. One gotcha: concept names vary by company Companies don't all use the same GAAP concept names to report the same thing. Apple reports revenue as RevenueFromContractWithCustomerExcludingAssessedTax . Older companies use Revenues . Some use SalesRevenueNet . If you just look up one concept name, you'll get blanks for most companies. The fix is a concept alias map: try each name in order, use the first one that has data. const CONCEPT_MAP : Record < string , string [] > = { revenue : [ ' Revenues ' , ' RevenueFromContractWithCustomerExcludingAssessedTax ' , ' RevenueFromContractWithCustomerIncludingAssessedTax ' , ' SalesRevenueNet ' , ' SalesRevenueGoodsNet ' , ], netIncome : [ ' NetIncomeLoss ' , ' NetIncomeLossAvailableToCommonStockholdersBasic ' , ' ProfitLoss ' , ], operatingCashFlow : [ ' NetCashProvidedByUsedInOperatingActivities ' , ' NetCashProvidedB

While working on a MERN project, I came across these indexes: transactionSchema . index ({ user : 1 , date : - 1 }); transactionSchema . index ({ user : 1 , type : - 1 }); transactionSchema . index ({ user : 1 , category : - 1 }); My first reaction was: "Why are we creating 3 different indexes for the same schema? Isn't one index enough?" At that time, my understanding was: "Indexes help MongoDB find records faster." Which is true, but it wasn't enough to explain why multiple indexes existed for the same collection. That simple doubt led me down a rabbit hole of learning about indexes, compound indexes, how MongoDB stores them, and the famous Prefix Rule. Here's what I learned. What is an Index? Imagine a collection with millions of transactions. db . transactions . find ({ user : " Aarthi " }); Without an index, MongoDB may need to inspect every document until it finds the matching records. This is called a Collection Scan . Think of it like searching for a chapter in a book without a table of contents. You'd have to flip through page after page until you find it. An index works like a book's table of contents. Instead of scanning every document, MongoDB can jump directly to the relevant records. Example: db . transactions . createIndex ({ user : 1 }); Now MongoDB can quickly locate all transactions belonging to a specific user. What is a Compound Index? A compound index contains multiple fields. Example: db . transactions . createIndex ({ user : 1 , date : - 1 }); This means MongoDB organizes the index by: user └── date Conceptually, it looks something like: Aarthi 2025-08-10 2025-08-09 2025-08-08 John 2025-08-10 2025-08-05 The data is first grouped by user , and within each user, it is ordered by date . Now queries like: db . transactions . find ({ user : " Aarthi " }). sort ({ date : - 1 }); become very efficient. MongoDB can jump directly to Aarthi's records and retrieve them in date order. The Prefix Rule: The Concept That Finally Made It Click Consider this i

PDF work shows up in dev life more often than we'd like to admit — exporting docs, compressing build artifacts, merging client deliverables, or converting a spec sheet someone sent as a scanned PDF into something you can actually search. Paid suites like Adobe Acrobat are overkill for most of these one-off tasks. Here are 10 free, no-signup tools that get the job done, ranked roughly by how often you'll reach for them. 1. ToolTiny — PDF to Word/Excel/PowerPoint ToolTiny converts PDFs into editable DOCX, XLSX, or PPTX files directly in the browser, alongside the usual merge/split/compress/watermark/password toolkit. No account, no watermark on output. What's actually useful for dev workflows: it handles presentation-style PDFs (think exported slide decks or design-heavy one-pagers) reasonably well — most converters flatten these into a single unreadable text blob, but ToolTiny keeps the layout intact while still giving you editable text. Good for the "client sent a PDF, I need it as a Word doc by EOD" scenario. 2. Smallpdf The OG in this space. Smallpdf's PDF-to-Word conversion is excellent at preserving layout — it renders the page as a background image and overlays editable text boxes at the correct coordinates, which is why it handles complex layouts better than most. Free tier caps you at 2 tasks/day though. 3. iLovePDF Similar feature set to Smallpdf, slightly more generous free tier. Their "Organize PDF" drag-and-drop page reordering is one of the smoother UX implementations out there if you need to quickly reshuffle a multi-doc PDF before sending it out. 4. PDF24 A German tool that's been around forever and quietly does everything — OCR, forms, signing, comparison. Less polished UI than the others but the OCR accuracy on scanned technical docs is genuinely strong. 5. Stirling-PDF If you want something self-hosted, Stirling-PDF is the open-source answer. It's a Docker container you spin up yourself, giving you a full PDF toolkit (split, merge, compress, OCR, wa

The Problem I Kept Seeing Over the past year working across multiple client teams on DevOps and pipeline work, I kept noticing the same thing. Developers storing GitHub PATs in Notepad. QA engineers keeping API keys in a text file on the desktop. DevOps folks with database passwords in a sticky note app. During screen shares — sprint reviews, debugging sessions, pair programming, recorded demos — those credentials were just sitting there. Visible to everyone in the call. Nobody said anything. It just kept happening. Why Existing Tools Didn't Fit I looked for something simple that solved this. Here's what I found and why none of it quite worked: Password managers (1Password, Bitwarden) Good tools. But they're built around cloud sync, browser extensions, and team sharing. For an individual developer who just wants somewhere safe to keep a PAT — overkill. Also: corporate IT policies often block installation of cloud-synced password managers on work machines. Secret managers (HashiCorp Vault, AWS Secrets Manager) These are infrastructure tools, not personal workflow tools. Setting up Vault for an individual developer's PAT collection is like using a forklift to move a chair. OS keystores (Windows Credential Manager, macOS Keychain) Actually decent for storage. But no UI built for this workflow, no copy-to-clipboard, and they don't solve the screen-exposure problem at all. The gap: Something simple, local, and designed around the moment of use — not just storage. So I Built Tokenly Tokenly is a local-only desktop credential vault. The core design principle is simple: Credential values are never shown on screen. You copy them to clipboard. That's the only way to use them. The clipboard auto-clears after 30 seconds. If you need to visually verify a value — press and hold a button. Release it, the value hides immediately. Not a toggle — a hold. Toggles get forgotten. Holds don't. Technical Decisions Worth Explaining Why Tauri over Electron Tauri uses the operating system's

AI can generate unit tests in seconds. But how do you know whether those tests are actually useful? Most teams still rely on code coverage and pass rates to evaluate their test suites. The problem is that a test can pass, increase coverage, and still provide little or no additional confidence. We've been seeing examples where AI-generated tests: Duplicate existing coverage Depend on system time or GUID generation Access files, network resources, or environment variables Use ineffective or unnecessary mocking Add maintenance cost without improving quality Today we launched Typemock Test Review, a tool that analyzes tests during execution and identifies duplicate, fragile, ineffective, and high-maintenance tests. Instead of looking only at source code, it combines runtime behavior, code coverage, dependency analysis, assertions, and mocking patterns to determine whether a test is actually contributing value. Some of the issues it can detect: Duplicate tests Hidden external dependencies Flaky test risks Unused or stale fakes Ineffective mocking Tests that increase maintenance without increasing confidence I'm curious how other teams are dealing with the explosion of AI-generated tests. Are you reviewing AI-generated tests differently from manually written tests? Have you found good ways to measure test quality beyond coverage and pass/fail metrics?