🌍 Dev.to

Your agent ran git. So did the repo

On October 2, Manifold Security published GitSpawn , and it is the kind of bug that is obvious the second you see it and invisible until then. Coding agents call git all the time. Not when you ask them to commit: in the background, to gather context. What changed, what branch you are on, what the diff looks like. And git reads its settings from the repository's own .git/config . Some of those settings name a program to run. The one Manifold used is core.fsmonitor , a hook git calls to ask ...
🌍 Dev.to

I built an open-weight-model agent that finds OSS issues worth your time

What I Built ClaimScout — a tool that tells you which open-source issues are actually worth your time. Point it at the repos you care about and it hands back a short list of issues that are genuinely yours to take : unclaimed, low-traffic, bounded bugs — not assigned work, not bot chores, not architecture epics dressed up as quick wins. Then, if you want, an open-weight model ranks that shortlist against your skills so the best first contribution floats to the top. Who would pay for...
🌍 Dev.to

Is 'Codex as an army' actually a trap for solo developers?

This came up in Chinese AI developer communities this week, and the energy around it was hard to ignore. A post celebrating OpenAI Codex drew 432 comments in a short window. The author described the feeling as "one person commanding a thousand troops." One commenter had shipped an automated video editing pipeline using a local Qwen model for frame tagging, DeepSeek for script rewriting, and a custom AI director layer for cut decisions, all built primarily through Codex. Another had automated the...
🌍 Dev.to

$0 to $1,869 a Month: What a Static Website Really Costs on Azure, by Size

"Static sites are basically free to host." That's mostly true, until it suddenly isn't. On Azure, the same pile of HTML, CSS and JavaScript can cost $0, $9, $69 or $1,869 a month . What changes the bill is how big the site is, and above all how much of it people download. I wanted to know exactly where each one of my sites would sit as it grows. So I priced five static sites, from a tiny portfolio to a site serving 20 TB a month, against every reasonable way to host them on Azure. ...
🌍 Dev.to

Your job feed is lying: detecting silent JSON API breakage (a stdlib-only GitHub Action)

I consume a lot of public JSON APIs, and the worst failure mode is the one that never throws: the feed still returns HTTP 200 but the shape has quietly changed. A field gets renamed. The item count drops from 800 to 12 overnight. A salary key disappears. An endpoint starts returning an empty array. None of it is an error — your HTTP client is happy — so a plain status-code health check says "all good" while your data is already broken. That is the gap jobfeed-watchdog (JFW) fills: a s...
🌍 Dev.to

Are 'Best AI Tools' Rankings Actually Hurting Your Workflow Decisions?

This came up in Chinese AI developer and marketing circles recently: a well-known AI business consultant published a 2026 ranked list of the best AI tools by category, and within days it was being screenshot, reshared, and acted on by people restructuring real production pipelines. Claude won writing. Higgsfield beat Sora and KlingAI for video generation. The list was clean, confident, and shareable. It was also already partially stale by the time most people saw it, and the comment section m...
🌍 Dev.to

How to Optimize Images Without Sacrificing Quality.

Images make websites more engaging, explain complex ideas, and help visitors understand products and services. However, large image files can slow down page loading and affect the user experience. The challenge is to reduce image file sizes without making images look blurry, pixelated, or unprofessional. In this guide, we'll explore practical ways to optimize website images while maintaining their visual quality. Why Is Image Optimization Important? Unoptimized images can increase pa...
🌍 Dev.to

免費的 Colab CLI A100/H100 真香 :)

前幾天 Google 官方宣布 Google AI Pro 方案納入 Colab 進階算力 ,讓訂閱者能直接在雲端使用 A100/H100 GPU。平常無聊跑些小模型時,我都會透過 AGY Cli 呼叫 google-colab-cli 來使用以前的 T4。既然這次 Google 這麼佛心給了 A100,我馬上決定直接從本機連上去。從認證、開立實例到載入模型,把這免費資源好好測過一遍。 釐清兩款 CLI 工具 PyPI 上有一個舊套件叫 colab-cli (版本 2.2.5),如果你跑 pip install colab-cli ,它會把環境裡的 click 降級到 7.1.2、 typer 降級到 0.1.1,接著就會發生一堆套件依賴衝突。看它的原始碼就知道,這只是幾年前包裝 pydrive 的上傳小工具,只能把本機的 .ipynb 同步到 Google Drive 再幫你開瀏覽器分頁,根本沒有遠端調用 GPU 的能力。 真正的官方工具是 Google 在 2026 年推出的 google-colab-cli ...
🌍 Dev.to

How to convert CGPA to percentage (and why the 9.5 rule isn't universal)

Many students need to convert a CGPA into a percentage for job applications, scholarship forms, or further study. The conversion looks simple, but the right formula depends on which body issued your grade sheet. The commonly cited formula For CBSE results, the formula most often quoted is: percentage = CGPA x 9.5 Worked example: a CGPA of 8.2 gives 8.2 x 9.5 = 77.9% The 9.5 rule isn't universal The 9.5 factor is a CBSE convention. Many universities publi...
🌍 Dev.to

1 in 5 AI-Suggested Packages Doesn't Exist — I Built a Scanner for That

1 in 5 AI-Suggested Packages Doesn't Exist — I Built a Scanner for That You ask an AI coding assistant for help. It writes the code and the requirements.txt . You pip install -r requirements.txt . And one of those package names? The AI made it up. Research presented at USENIX Security 2025 measured this at scale: across 2.23 million package references produced by 16 large language models, 19.7% pointed to packages that were never published . Attackers know this too — they regist...
🌍 Dev.to

A task feed should distinguish a reference from a result

When a task library exposes an API, downstream readers can lose qualifications that were obvious on the original page. A title such as “Prepare a school supply cart” can become a capability claim if the consuming application drops the source and evidence label. I maintain Muse Task Hub , an independent, unofficial guide. Its public feed is at the /api/tasks path on the guide’s domain. The records include taskId, title, summary, category, evidenceType, url, sources, version, contentVersion,...
BriefBill: Turning Vague Client Briefs into Clear Freelance Quotes with Local Gemma 3
🌍 Dev.to

BriefBill: Turning Vague Client Briefs into Clear Freelance Quotes with Local Gemma 3

This is a submission for the Hacktoberfest Open-Source AI Challenge: Side Hustle What I Built BriefBill turns a messy client message into a clear freelance proposal with transparent pricing. Freelancers get briefs like "need a website, maybe some posts too, budget is tight." Quoting from that is risky. You either undercharge or agree to work nobody defined. BriefBill helps in three ways: Reads the brief with Gemma 3 (4B), running locally through Ollama Calcu...
Android developer verification: four countries first, global in 2027
🌍 Dev.to

Android developer verification: four countries first, global in 2027

If your Android app reaches people outside Google Play, the install path is changing for some of them. Enforcement began on 30 September 2026 in Brazil, Indonesia, Singapore and Thailand, and Google says the global rollout follows in 2027. What Google's guide says The developer verification guide was last updated on 18 August 2026. It says: "Starting September 30, 2026, these new developer verification protections go live for users in Brazil, Indonesia, Singapore, and Thailand." ...
Invoice AI
🌍 Dev.to

Invoice AI

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle What I Built InvoicePilot AI is a complete, production-ready, open-source AI application built for freelancers, solo founders, and small businesses to organize invoices and receipts without manual entry, expensive SaaS subscriptions, or sending private financial records to cloud AI vendors. The Problem It Solves Solo operators and independent contractors waste hours every ...
InvoiceGenie: The 100% Private, Local AI Financial Assistant & Receipt Sorter 🧾
🌍 Dev.to

InvoiceGenie: The 100% Private, Local AI Financial Assistant & Receipt Sorter 🧾

What I Built Meet InvoiceGenie—a local, offline AI-powered financial assistant and receipt sorter designed specifically for freelancers, independent contractors, and small business owners. The Problem : Freelancers juggle multiple clients, irregular invoices, software subscriptions, and messy expense receipts. Manual bookkeeping is tedious, and commercial cloud-based finance tools either charge expensive monthly subscriptions or pose serious data privacy risks when handling sensitive cl...
How Unicode Text Generators Create Aesthetic Fonts
🌍 Dev.to

How Unicode Text Generators Create Aesthetic Fonts

Aesthetic text generators can make ordinary words look cursive, bold, gothic, vaporwave, circled, or decorative. At first glance, this can look like a website is applying different fonts. In many cases, that is not what is happening. Instead, the generator is replacing normal characters with visually similar Unicode characters. Unicode Characters vs CSS Fonts A normal CSS font changes how the same underlying text is displayed. For example: font-family: Georgia, s...
🌍 Dev.to

What a Logseq page looks like after export to .docx, no Pandoc (real app, real file)

Disclosure The author works at Quillfern, an AI-assisted studio, and the plugin shown here is the studio's, which is a potential conflict of interest. The plugin also has an optional paid tier; this article covers only the free version. This article was written with AI assistance. What I tested A small page in Logseq OG 1.0.0 (the Markdown file-graph app, run on Linux under a virtual display): a nested outline with a link, two task blocks (one TODO, one DONE) and a three-colu...
Precise location in Android apps: the location button comes first
🌍 Dev.to

Precise location in Android apps: the location button comes first

Many Android apps ask for precise location by default. Google's preview policy names the location button as the recommended minimum for precise location, and it takes effect on 27 January 2027. What Google's preview says The Play Console help page is a preview. It says: "This article previews changes from our April 2026 policy updates." For location, the page says: "We're introducing the location button as the recommended minimum scope for precise location." The effective date is...
🌍 Dev.to

An Agent Patch May Not Run Until the Fixture Digest Is Signed

An agent patch is not ready for a property suite because a model emitted it. It is ready only when three records exist and play different roles: a provenance note, a fixture digest signed outside the drafter, and a property class that limits what a pass or fail may mean. Miss any one of them and the run should not start. A flake freeze is a later ledger. It cannot repair a missing signature. That admission order is the strategy. The steps below enforce it without treating a cheaper draft, or ...
COCOS 4 Is Open Source — But Is It Time to Leave Cocos Creator 3.8?
🌍 Dev.to

COCOS 4 Is Open Source — But Is It Time to Leave Cocos Creator 3.8?

description: "COCOS 4 went fully open source under MIT in early 2026. I run a Cocos Creator 3.8 mini-game in production — here's my honest call on whether to migrate, and what actually changed." Cross-post note: This is the original article from my blog, Indie Game Lab . Read it there for the properly styled version, or keep reading here — it's identical. TL;DR: I run a Cocos Creator 3.8 WeChat mini-game in production. COCOS 4 — fully open-sourced under MIT in early 2026 — is the...
🌍 Dev.to

Record Observed Outputs Before One Safe Refactor

A messy function should not be rewritten on day one. Lock today's outputs with characterization tests before any edit. The smallest safe change is the edit those tests still accept. This workflow is a proposal with unexecuted examples. It does not report a measured production run. Use the workflow to separate observation from cleanup. Why a wide rewrite fails review A large rewrite hides several behavior changes in one diff. Reviewers then cannot name the line that moved an ou...
Text and Images in One Embedding Space on a CPU: What I Measured Building a Semantic grep
🌍 Dev.to

Text and Images in One Embedding Space on a CPU: What I Measured Building a Semantic grep

When I look for an old note or a screenshot, I often remember what it was about but not a single word in it. ripgrep is instant if you remember the word. That doesn't help when I remember the topic but not the wording, like "the meeting note where we decided to cut the budget" or "that dialog saying the disk was full". So I built anygrep , a CLI that puts notes, PDF pages and images into one local index and searches them by meaning. It runs one model, Google's EmbeddingGemma 2, as an int8 ON...
How Shallow and Deep Copies Work in Python
🌍 HackerNoon

How Shallow and Deep Copies Work in Python

I copied a Python dictionary, but the original still changed. Here's how shallow copies, shared references, and deep copies actually work.