SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingDEV Community ·

Dantiva Creator Shares Hard Lessons From Building an AI Video App for Russian Users

A solo developer built Dantiva, an AI-powered tool that converts photos into short video clips using Google's Veo and Gemini models, accessible via a website, Telegram bot, and Telegram Mini App. The product targets Russian-speaking users who face barriers accessing Western creative tools, offering ruble payments and receipts through a familiar Telegram interface. While the core AI integration was straightforward, the developer says the bulk of the effort went into building surrounding infrastructure — accounts, token ledgers, templates, legal pages, and an admin panel. A key technical challenge was unifying identity and token balances across three separate entry points, a problem that took roughly a month to solve and still surfaced a UI bug on September 19 that incorrectly showed unlimited-access users a paid balance. Paid subscriptions launched on September 18, and the developer notes that actual purchase data over the coming weeks will be the real test of whether the market hypothesis holds.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Speed Alone Won't Save Your RAG System: Why Embedding Evaluation Matters

A developer guide published on DEV Community warns that retrieval-augmented generation (RAG) pipelines can return results in milliseconds and still surface wrong or irrelevant answers. The core issue lies not in vector generation speed but in whether embeddings rank the most relevant documents closest to a given query. Three information-retrieval metrics — Recall@k, MRR, and NDCG — are highlighted as practical tools for measuring retrieval quality. The guide also flags real-world challenges such as multilingual queries, code-switched text, and domain-specific terminology absent from a model's training data. It recommends building a benchmark test set from actual user queries to objectively compare embedding models and decide when fine-tuning is warranted.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds MCP Server to Feed AI Agents Design Data from Awwwards

A developer has created an open-source MCP (Model Context Protocol) server that connects AI agents to Awwwards.com, a platform showcasing award-winning website designs. The tool effectively transforms the design showcase site into a structured, agent-readable library. It offers seven tools operating over stdio and is built using TypeScript with the official MCP SDK. Notably, the project has zero native dependencies, keeping it lightweight and easy to integrate. The source code has been made publicly available on GitHub.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Claude Code Silently Compacts Context at 160K Tokens, Dropping Critical Session Memory

Claude Code, Anthropic's AI coding agent, automatically triggers context summarization when a session reaches approximately 160,000 tokens — about 80% of its 200,000-token limit. During this process, the middle portion of the conversation is condensed into a prose summary and original messages are discarded without any visible warning to the developer. The summarization preserves recent exchanges and the system prompt but loses detailed reasoning chains, rejected approaches, and edge cases discovered earlier in the session. This silent compaction can cause the agent to produce code that violates constraints it had previously acknowledged, leading to subtle but costly errors in production. Developers are advised to externalize critical constraints and decisions into persistent artifacts — such as decision logs or constraint manifests — to maintain continuity across compaction boundaries.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds Offline AI Tool That Converts Static HTML Sites Into WordPress Themes

A developer has created a local AI-powered tool that automatically converts static HTML websites into installable WordPress themes without requiring cloud services, API keys, or paid subscriptions. The tool uses Gemma 2B, a lightweight local language model, combined with Retrieval-Augmented Generation (RAG) to detect and map HTML components such as headers, footers, and navigation into standard WordPress template files. DOM parsing and heuristics work alongside the AI to handle varied HTML structures, while the original CSS is preserved to maintain responsive layouts across devices. Users point the tool at a folder of HTML files, and it outputs a ready-to-install ZIP containing all necessary WordPress theme files including functions.php, header.php, and style.css. The tool runs entirely on the user's machine after a one-time model download, making it a cost-free alternative to manual theme conversion or cloud-based solutions.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Microsoft Director Called AI Scraping Largest Labor Theft in History, Court Docs Reveal

Internal statements from Microsoft and OpenAI executives have surfaced through legal briefs filed in the New York Times lawsuit against the two companies. A Microsoft director reportedly described AI scraping of online content as the largest theft of labor in human history. Separately, an OpenAI executive allegedly acknowledged that ChatGPT poses an existential threat to publishers. The revelations emerged not from whistleblowers but from court documents submitted as part of the ongoing litigation. The disclosures add significant weight to the NYT's legal case and highlight tensions within the AI industry over the use of copyrighted material for training models.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

10 Essential Linux Terminal Commands Every Beginner Should Know

A beginner-focused guide published on DEV Community outlines ten fundamental Linux terminal commands for new users. The list covers file navigation and management tools such as ls, pwd, and cp, which help users browse directories and copy files. Utility commands like df -h, cat, and clear allow users to check disk space, view file contents, and tidy the terminal screen. Additional commands including whoami and history let users identify the active account and review past terminal activity. Together, these commands form a practical foundation for anyone starting out with the Linux command line.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

PaperCut Flaws Exploited at Scale in 2026, But Scan Counts Mislead on True Exposure

Two critical vulnerabilities in PaperCut NG and MF print management software — CVE-2026-81578 (CVSS 9.8) and CVE-2026-82078 (CVSS 9.1) — were actively exploited in September 2026, enabling authentication bypass and remote code execution. Internet scan data from ZoomEye returned over 1.2 million results for PaperCut-related queries, but analysts warn these figures do not represent the actual count of vulnerable servers. The high numbers reflect how broadly the PaperCut name appears across web content, including pages that merely reference or link to the platform rather than host it. More precise queries, such as those targeting PaperCut NG specifically or the CVE identifier, returned zero results, exposing gaps in the scan index rather than confirming the absence of vulnerable systems. Security teams are advised to scope exposure through internal asset inventories, verify software versions against patched releases, and restrict public internet access to print management servers.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

New AI model 'Jev' cuts agent loop latency by skipping text generation for decisions

A common inefficiency in AI agent loops is that language models generate verbose prose just to produce a simple decision — such as which element to click — wasting seconds and hundreds of tokens per step. TypeSafe released a model called Jev four days ago that addresses this by accepting typed questions (yes/no, multiple choice, ratings) alongside a state input and returning structured answers with confidence scores in a single forward pass, without generating a token stream. A developer built an open-source plugin called jev-use that routes decision-only steps to Jev while keeping text-generation tasks with a standard LLM like Claude or Codex. Multiple questions about the same state can be batched into one call, keeping latency flat regardless of question count. Currently Jev is API-only with no self-hosted option, meaning state data such as page DOMs or command outputs must leave the user's machine to be processed.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How a word-puzzle game replaced flat coin rewards with difficulty-scaled payouts

A developer building a word-block puzzle engine overhauled its in-game coin economy after discovering that a flat 20-coin reward per level caused players to grind the easiest stages repeatedly. The new system ties payouts to a metric called 'decision load' — the total number of meaningful choices a board forces on a player — which ranges from 7.5 on easy levels to 96.7 on the hardest, a 13-times spread. Star-rating bonuses are calculated as incremental deltas against the best tier a player has previously completed for that level, preventing lucrative replays of already-mastered content. Rewarded video ads offer a multiplier rather than a flat bonus, so the reward naturally scales with level difficulty and discourages easy-level farming. A capped daily login chest replaced an earlier uncapped version that had ballooned to paying out more coins than three level completions combined by its eighth weekly cycle.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Study Tests Btrfs, ZFS, and Bcachefs Under Real-World Workloads Benchmarks Miss

A benchmark study published at bartosz.fenski.pl evaluates three modern Linux filesystems — Btrfs, ZFS, and Bcachefs — under workloads that traditional benchmarks typically overlook. The analysis aims to fill a gap left by classic filesystem benchmarks, which often fail to reflect real-world usage patterns. The study compares performance across these filesystems in scenarios more representative of everyday and production environments. The article attracted attention on Hacker News, gathering points and community discussion. The findings offer practical insights for users and administrators choosing between these advanced copy-on-write filesystems.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

RAG Pipelines Face 'Knowledge Injection' Attacks That Hijack LLM Responses

Retrieval-Augmented Generation (RAG), widely used to ground large language models in private or real-time data, contains a critical vulnerability in its core architecture. Attackers can plant malicious instructions inside documents, web pages, or tool outputs that the RAG pipeline retrieves, causing the LLM to treat adversarial content as authoritative. This technique, called knowledge injection or indirect prompt injection, differs from direct prompt injection because the attacker targets the data source rather than the user's query. A detailed engineering analysis published on DEV Community outlines why standard RAG implementations are inherently fragile due to implicit trust in retrieved content. The article proposes a layered defense strategy covering source isolation, semantic threat detection, and dual-model orchestration to harden RAG systems against such attacks.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

CISA Flags Two SonicWall SMA 1000 Flaws as Actively Exploited, Gives Agencies 3 Days to Patch

CISA added two SonicWall SMA 1000 vulnerabilities — a pre-authentication SSRF flaw (CVE-2026-83548) and an OS command injection flaw (CVE-2026-83549) — to its Known Exploited Vulnerabilities catalog on September 2, 2026, with a federal remediation deadline of September 5. SonicWall responded the same day by publishing advisory SNWLID-2026-0016 alongside patched firmware versions 12.4.3-03526 and 12.5.0-02952. A ZoomEye scan conducted on September 18, 2026, identified only 7 records matching the precise SMA product fingerprint, reflecting that the appliance's management interface is not intended for public exposure. Security researchers caution that broader vendor-name searches returning millions of SonicWall-branded assets do not accurately represent SMA 1000 exposure and should not be conflated with the specific vulnerability surface. Organizations are advised to query their own perimeters using the product-specific fingerprint and treat any internet-reachable SMA 1000 portal as potentially vulnerable until patched versions are confirmed.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Three MCP Servers That Extend Claude Desktop With Files, GitHub, and Postgres

A developer has shared three Model Context Protocol (MCP) servers they actively use with Claude Desktop to enhance productivity. The filesystem server allows Claude to directly read project files, eliminating the need to manually copy-paste code into the chat. The GitHub server enables Claude to browse repositories, review pull requests, and summarize recent code changes. A PostgreSQL server grants Claude direct database query access without leaving the conversation. Each server can be installed via the mcp-hub CLI in roughly two minutes, though users are cautioned to restrict filesystem paths carefully to avoid unintended disk-wide access.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer open-sources SVG animation engine that draws strokes, fills, and labels in sequence

A developer has extracted and open-sourced an SVG animation library called @penwell/draw-on after it grew complex enough within a personal project to stand on its own. The library animates SVG diagrams by progressively drawing strokes first, then revealing fills, and finally displaying labels, offering a more natural reveal than simple fade-ins. Unlike its predecessor Vivus, draw-on supports manual progress control, allowing the animation to be tied to scroll position, custom timelines, or other systems. It also includes a React API while keeping the core package framework-agnostic, and can render animations directly to WebM video in the browser without requiring a server. The package is available on npm under the MIT license at version 0.1.4, and the developer is seeking feedback from users working with SVGs from tools like Mermaid, Figma, and Excalidraw.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How 'Share of Model' Is Redefining Visibility in the Age of AI Search

As AI assistants like ChatGPT, Claude, and Gemini increasingly answer user queries directly, traditional SEO metrics such as rankings and click-through rates are no longer sufficient measures of online visibility. A new concept called 'Share of Model' tracks how often a brand, product, or entity appears in relevant AI-generated responses across these platforms. Unlike conventional search, language models weigh source consistency, credibility, and clarity before including any entity in an answer, meaning vague or contradictory information can effectively penalize a brand's chances of being mentioned. Analysts refer to this risk as an 'Uncertainty Penalty,' where ambiguous or outdated information makes it harder for AI systems to confidently recommend an entity. To improve their Share of Model, organizations are advised to maintain accurate, well-structured, and corroborated information online, and to track AI responses using fixed prompt sets rather than treating results as precise, comparable metrics.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds Console Snake and Ladder Game in Go to Teach Clean Code Design

A developer published a console-based Snake and Ladder game written in Go, sharing it as a practical example of clean software design principles. The project features a 100-square board with snakes, ladders, and three player tokens that take turns rolling a six-sided die. Each component — Board, Dice, Coin, Snake, and Ladder — is built as a separate type with a single responsibility, following common object-oriented design patterns. Snakes and ladders share a common interface called Jumps, allowing the game logic to handle both without distinguishing between them. The project is intended to help developers understand concepts such as facades, polymorphism, and interface-based design in Go.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer loses IP warm-up after 24% bounce rate despite SMTP 250 OK verification

A developer running a cold email campaign on March 14 sent 50 messages to leads that had all passed basic SMTP verification, only to see 12 bounce — a 24% failure rate. The core issue is that an SMTP 250 OK response confirms a server accepted the envelope, but does not guarantee the mailbox exists or is actively used. Using a third-party email validation API, the developer found that addresses like test@gmail.com, while syntactically valid and SMTP-friendly, are flagged as role addresses with reduced deliverability scores. The API returned additional signals — including null SMTP verification status, a deliverability score of 75, and breach history — that a raw SMTP handshake cannot provide. The incident cost the developer a fresh IP warm-up, roughly three hours of list cleanup, and required explaining the bounce spike to a client.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

503 vs 504 Errors Explained: How nginx, AWS ALB, and Cloudflare Differ

HTTP 503 and 504 errors signal fundamentally different problems — availability versus latency — and diagnosing them correctly depends on which infrastructure layer is involved. In nginx, a 504 is triggered when the gap between successive reads from an upstream exceeds proxy_read_timeout, while 503s typically come from rate limiting or the application layer, never from a slow upstream. AWS ALB returns a 503 when no healthy targets exist in a target group, and a 504 when a healthy target fails to respond within the configurable idle timeout, which defaults to 60 seconds. Cloudflare uses its own 524 code when an origin connects via TCP but returns no HTTP response within roughly 100 seconds, reserving 503 for requests blocked by its own WAF or rate-limiting rules at the edge. Engineers troubleshooting these errors are advised to check target health and recent deployments for 503s, and to profile the slow operation itself before raising any timeout thresholds for 504s.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Releases 150 GB Open-Source Dataset for Central Asian Languages and Code

A developer has published a nearly 150 GB multilingual dataset on Hugging Face to address the severe shortage of open-source training data for Central Asian languages, including Kyrgyz, Kazakh, Uzbek, and Tajik. The dataset also includes 20 GB of source code in Python, C++, Rust, and Go, making it one of the few resources combining low-resource languages with technical corpora. Building the dataset took weeks of collecting, parsing, and cleaning data, while compressing the archive to around 27.7 GB caused repeated out-of-memory crashes and took roughly 10 hours to complete. The project is released under the CC BY 4.0 license, allowing free use for research, benchmarking, and commercial purposes. It is intended to support tasks such as code LLM fine-tuning, technical translation, and continual pre-training of lightweight language models tailored to Central Asian contexts.

0 comments Read more at DEV Community
← NewerPage 1115 of 3219Older →