SShortSingh.

Programming

0
ProgrammingDEV Community ·

GitSpawn Flaw Let Malicious Repos Run Code When Opened in AI Coding Tools

Manifold Security researchers discovered a vulnerability class called GitSpawn affecting seven popular AI coding agents, including Claude Code, Codex, Cursor, and Grok Build. The flaw exploited Git's legitimate core.fsmonitor configuration option, which can point to an external executable program. When developers simply opened a project folder, the coding tool would run background Git commands like git status, inadvertently triggering attacker-controlled code without any user approval prompt. Eight related issues were identified across the affected tools, with exploitation requiring no prompt injection or sandbox escape — just opening a repository was enough. The findings highlight a broader security concern: modern development environments have quietly blurred the line between browsing a codebase and executing it.

0
ProgrammingDEV Community ·

Developer Builds Monadic ML Framework to Bridge Research and Production Gap

A developer and ML practitioner has built a Python-based machine learning framework designed to streamline the full ML lifecycle, from research to production deployment. The framework draws on monadic design principles from functional programming to enable composable, reproducible, and production-safe pipelines. It centers on two core abstractions: a DataPod object that carries data and shared state, and Transformer modules that encapsulate behavior, learn during training, and store fitted parameters for later reuse. Together, these components allow ML pipelines built during experimentation to be replayed reliably in production without rewriting code. The approach aims to eliminate common pain points such as library fragmentation, schema mismatches, and brittle pipeline logic that plague typical ML projects.

0
ProgrammingDEV Community ·

Google Play Closed vs Open Testing: What New App Developers Must Know

Google Play Console offers multiple release tracks, and understanding the difference between closed and open testing is essential for a smooth app launch. Personal developer accounts created on or after November 13, 2023 must complete closed testing — requiring at least 12 opted-in testers for 14 consecutive days — before they can apply for production access. Open testing, designed for large-scale public beta programs, does not fulfil this mandatory requirement and cannot be used as a substitute. Developers on organization accounts or older personal accounts predating November 13, 2023 are exempt from the 14-day rule and may skip directly to open testing or production. Once the closed testing gate is cleared, developers can still run open testing alongside a live production build to safely evaluate major updates on real devices.

0
ProgrammingDEV Community ·

Browser-Side File Processing Can Handle Conversions Without Uploading Your Data

Many online file tools require users to upload their files to remote servers, but modern browsers can handle common tasks like image conversion, compression, and PDF operations entirely on the user's device. Using technologies such as JavaScript, WebAssembly, and Web APIs, files can be read, processed, and downloaded locally without ever leaving the browser. This approach offers privacy benefits, as sensitive documents and photos are never transmitted to external servers, along with faster processing since no upload wait time is involved. However, local browser processing has limitations, including high memory usage for large files and performance dependence on the user's hardware. Developers building file-handling tools are encouraged to evaluate whether server-side uploads are truly necessary before defaulting to that approach.

0
ProgrammingDEV Community ·

Docker Engine 29 silently ignores container disk quotas under new default backend

Docker Engine 29 changed its default image store from the classic overlay2 graphdriver to the containerd snapshotter for new installations. A hands-on comparison revealed that the --storage-opt size= flag, meant to cap a container's writable layer, is silently ignored under the new default backend. While overlay2 correctly refuses to start a container when the required XFS filesystem with pquota is absent, the containerd snapshotter accepts the flag, starts the container, and allows writes far beyond the specified limit with no error or warning. Making matters worse, docker inspect still reports the quota as active, meaning compliance checks and audit tools would falsely indicate the limit is enforced. The issue appears to stem from the containerd image store path lacking the filesystem-support validation that the graphdriver path performs, and the behavior change was introduced without any CLI-level notice to users.

0
ProgrammingDEV Community ·

NovelAI V5 Models Work in ComfyUI but Anlas Balance Error Needs a Manual Fix

Both nai-diffusion-5-full and nai-diffusion-5-curated model IDs function correctly through the ComfyUI_NAIDGenerator node, despite a README caveat stating they had not been verified against the live API. A widely reported 'Error fetching Anlas 400' is unrelated to the models themselves and stems from NovelAI having moved its user endpoints from api.novelai.net to image.novelai.net around the V5 launch. The fix requires updating USER_API_BASE_URL in two places within nodes.py to point to the new host. An additional complication arises because the new endpoint returns trainingStepsLeft as a nested object rather than an integer, causing a TypeError when the node attempts arithmetic on the balance. On an Opus-tier account, generating a standard 832×1216 image at 28 steps consumed zero Anlas, confirming the generation pipeline itself works as expected.

0
ProgrammingDEV Community ·

Experiment Shows AI Models Default to Negative Verdicts Without Real-World Evidence

A developer ran a controlled blind experiment across five AI model families — Grok, DeepSeek, GPT, Gemini, and Claude — asking each to argue whether AI makes humanity intellectually stronger or weaker. All ten runs, across two isolated arms each, unanimously concluded 'weaker,' drawing on the same well-documented research around cognitive offloading and memory decline. To investigate the cause, the developer ran a follow-up test with Gemini using four conditions, including one where the model was given a real three-month record of an individual's AI-assisted work, complete with failures and no instructed conclusion. That single change — introducing actual evidence — flipped the verdict to 'stronger' in both draws, while reworded or reframed questions without evidence still returned negative results. Analysis of the models' reasoning traces revealed they had defaulted to the 'weaker' position largely because the atrophy-focused literature is older, larger, and more citable, exposing how AI models can mistake bibliographic abundance for truth.

0
ProgrammingHacker News ·

JetKVM Launches Mini: A Compact Remote Access Device for Servers

JetKVM has announced the JetKVM Mini, a new compact version of its KVM-over-IP hardware device. The product allows users to remotely access and control computers or servers at the BIOS level without needing a physical presence. The announcement was shared on the JetKVM blog and gained traction on Hacker News with community discussion. The Mini appears to be a smaller form factor iteration of the existing JetKVM lineup, targeting users who need lightweight remote management solutions.

0
ProgrammingDEV Community ·

15 Ways Google Play Subscriptions Silently Fail and How to Fix Them

A developer running a self-managed Google Play Billing backend has documented 15 distinct failure modes that can quietly break subscription access without obvious errors. The issues range from double-processing renewals due to Pub/Sub redelivery, to users retaining access after cancellation or losing it during legitimate grace periods. Other problems include unacknowledged purchases being auto-refunded by Google after three days, upgrade flows leaving users with duplicate tier access, and real-time notifications silently stopping due to lost publisher permissions. Each failure mode corresponds to a specific rule in Billing Doctor, a free diagnostic tool the author built to trace where a subscription lifecycle broke down using Google's own documented billing behavior.

0
ProgrammingDEV Community ·

How to Build a Thread-Safe BlockingQueue in Java From Scratch

Implementing a custom BlockingQueue is a common Low-Level Design interview challenge at companies like Apple and Amazon, testing a candidate's grasp of thread coordination and JVM internals. A naive implementation using synchronized blocks with notifyAll() causes the 'thundering herd' problem, while using if instead of while leaves the queue vulnerable to spurious wakeups. The recommended approach uses a single ReentrantLock paired with two separate Condition instances — notFull and notEmpty — to isolate producer and consumer wait states. This design ensures that only the relevant thread is woken via signal(), reducing unnecessary context switching. Key implementation rules include always guarding await() calls with a while loop and using try-finally blocks to guarantee lock release even when exceptions occur.

0
ProgrammingHacker News ·

American Medical Association Formally Classifies Nearsightedness as a Disease

The American Medical Association (AMA) has endorsed the formal classification of myopia, commonly known as nearsightedness, as a disease. This decision marks a significant shift in how the condition is medically recognized in the United States. The reclassification was announced through the American Academy of Ophthalmology. Such a change could have implications for how myopia is treated, researched, and covered by insurance going forward.

0
ProgrammingDEV Community ·

WCAG 2.2 CSS Lint Tool Flags Accessibility Gaps Older Tools Miss

A CSS linting tool designed around WCAG 2.2 identified 14 accessibility findings across 61 lines of a typical frontend stylesheet, including three criteria introduced after WCAG 2.1. The new criteria — Target Size Minimum (2.5.8), Focus Not Obscured (2.4.11), and Focus Appearance (2.4.13) — are largely determined by CSS rules yet are invisible to linters or AI tools trained on older WCAG 2.1 standards. Common violations found included icon buttons under the required 24px touch target, sticky headers obscuring focused elements, and focus outlines that were too thin or suppressed entirely. Unlike browser-based tools such as Axe or Lighthouse, a stylesheet linter inspects source code directly, catching issues before a page route even exists. The tool's release coincides with the European Accessibility Act coming into force on 28 June 2025, making early detection increasingly important to avoid costly third-party audits.

0
ProgrammingDEV Community ·

Why 'Likes' Are a Poor Measure of Quality for Online Learning Platforms

A growing argument in educational technology suggests that the standard 'like' metric on learning platforms fails to capture why viewers actually engage with content. A single like can reflect vastly different motivations — a beginner finally grasping a concept, an advanced learner revising, or a viewer simply preferring content in their native language. This ambiguity creates a feedback loop where high engagement is mistaken for high teaching quality, causing recommendation algorithms to push popular content to all learners regardless of their level or needs. The problem is especially pronounced on global platforms, where language accessibility can drive engagement independently of instructional quality. Researchers and developers are exploring weighted engagement models that factor in learner level, language preference, topic familiarity, and demonstrated learning outcomes to produce more meaningful quality signals.

0
ProgrammingDEV Community ·

How Semantic Search Can Be Added to Any App in Three Simple Steps

A developer tutorial published on DEV Community explains how to upgrade a basic keyword-based search into AI-powered semantic search using just three core changes. The approach involves adding a vector column to a database table, generating embeddings via an OpenAI model when a record is created, and comparing vector distances at query time. The tutorial uses a Next.js todo app connected to PostgreSQL as a practical example, demonstrating how naive substring search fails to surface conceptually related results. The matching itself is handled by pgvector, a PostgreSQL extension that enables vector storage and distance-based querying — meaning no model inference occurs at search time. Both the original and upgraded versions of the app are available on GitHub, requiring only a pgvector-enabled Postgres instance and an OpenAI API key to run.

0
ProgrammingDEV Community ·

Reviactyl: Open-Source Game Server Management Panel Built on Modern Stack

Reviactyl is a newly launched open-source game server management panel designed to give users full control over their own hosting infrastructure. Built with Laravel, React, FilamentPHP, Vite, and Go, it allows individuals and hosting businesses to manage game servers through a modern web interface. Game servers run inside isolated Docker containers, ensuring workload separation and better resource control. The panel is self-hosted, meaning users bring their own VPS or dedicated server rather than relying on a third-party managed platform. Released under the MIT License, Reviactyl is free to use, modify, and extend, and also offers a developer-facing Extensions API for custom integrations.

0
ProgrammingDEV Community ·

NocoBase Bug: Filtered Bulk Updates Ignore Conditions, Risk Data Corruption

A code-level investigation into NocoBase 2.2.7 has revealed that bulk update operations do not carry filter conditions into the final SQL UPDATE statement. Instead, the platform first selects matching rows using the filter, then updates those rows by primary key alone — meaning rows that no longer meet the original condition at write time still get updated. The flaw was first reported on the official NocoBase forum on September 4th, with staff acknowledging they were investigating but no fix issued since. Testing on PostgreSQL 16 confirmed the behavior, with SQL logs showing the WHERE clause in the UPDATE containing only primary keys rather than the original filter. The vulnerability is rooted in the repository.ts update method, affects both available code paths, and poses a concurrency risk where stale or changed rows can be incorrectly overwritten.

0
ProgrammingDEV Community ·

Developer Releases V2 of Python Expense Tracker with Edit and Input Validation

A developer has released Version 2 of a personal expense tracker built using Python, OOP principles, SQLite, and SQL. The update adds an edit feature that allows users to modify existing expense details such as amount, category, description, and date. V2 also introduces comprehensive input validation to handle invalid menu choices, incorrect amounts, empty descriptions, and impossible dates like February 30. The project follows a layered architecture separating CLI flow, application logic, and database operations across three components. The developer noted that a key takeaway from this version was learning to define a clear scope and recognizing when a version is complete rather than continuously adding features.

0
ProgrammingDEV Community ·

Vibe Coding Is Fine for Fun, But a Risk in High-Stakes Software

A software engineer with ten years of experience argues that vibe coding — using AI to generate code without reviewing or understanding it — is acceptable for personal projects but dangerous when applied to systems handling sensitive data or financial transactions. The author distinguishes vibe coding from AI-assisted development, where the engineer still reads, reviews, and interrogates the generated code. The piece contends that labeling prompt-only, no-review development as 'engineering' is misleading, since engineering involves understanding, debugging, and accountability. The author warns that deploying unreviewed AI-generated code in production environments risks user data and security, and reflects a lack of curiosity about how software actually works. Until AI tools are reliable enough to work unsupervised, the author urges developers to stay in control and treat AI output with critical scrutiny.

0
ProgrammingDEV Community ·

Developer Builds Mac App That Replays Last 2 Minutes of Audio With a Hotkey

A developer created a macOS menu bar app called 'Huh?' after repeatedly missing spoken details — such as URLs and names — during Zoom calls and online lectures. The app stores a rolling audio buffer of the last 30 seconds to 2 minutes in RAM and replays it instantly when the user presses a keyboard shortcut. Users can slow playback to 0.75x or 0.5x speed without pitch distortion, and an on-device transcription feature helps identify missed words or links. No audio is written to disk or uploaded to any server unless the user explicitly saves or exports a clip. The project began after the developer shared a short clip on X, where several users expressed willingness to pay around $5 for the tool.

← NewerPage 1398 of 3334Older →