SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingHacker News ·

Anthropic Explores Using Claude to Shrink the Attack Surface of Software Code

Anthropic has published research on applying its Claude AI model to reduce the attack surface of software code. The work examines how the model might help identify and remove code that could be exploited by attackers. The post is shared on Hacker News, where it has drawn modest early attention. The source material available does not provide details on specific methods, results, or benchmarks.

0 comments Read more at Hacker News
0
ProgrammingHacker News ·

C# Compiled to WebAssembly Now Produces Smaller Binaries Than Rust

A GitHub project called NetWasm reports that C# code compiled to WebAssembly now yields smaller output than comparable Rust builds. The claim was shared on Hacker News, where the post gathered 25 points and 3 comments. The available material includes only the project link and discussion metadata, so the specific benchmark methodology and size figures are not detailed here. The result suggests progress in .NET's WebAssembly tooling relative to Rust's typical output sizes. Readers should consult the repository for the underlying measurements.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

Daily app budgets and a visible override counter outperformed blocking for one writer's phone habits

A developer who struggled with full blocks, accountability partners, and cold-turkey attempts over 2024 found each method removed access without addressing the underlying reasons for phone use. Their approach instead assigns per-app daily time budgets, with messaging and maps left unlimited, and prompts once at the limit to either continue or do an intended task first. Each override is recorded in a visible counter without punishment or streaks. Logging from 1 to 21 October reportedly cut late-night phone use from 90 to 12 minutes per night, daily opens from 41 to 19, and weekly overrides from 23 to 6. The writer has released the tool, ChoiceField, as a free iOS app and suggests a week of manual logging as a starting point.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

LLM API 장애에도 서비스가 멈추지 않도록 하는 6가지 설계 원칙

LLM API를 제품에 연결하는 일은 개발 단계보다 운영 단계에서 훨씬 까다로운데, 응답 지연, 네트워크 오류, Rate Limit, 서버 오류 등 다양한 실패가 언제든 발생할 수 있다. 이 글은 이런 실패를 전제로 한 호출 구조를 위해 여섯 가지 원칙을 제시한다. 핵심은 요청 목적에 맞는 타임아웃 설정, 오류 유형에 따른 선별적 재시도, 지수 백오프를 통한 부하 분산이다. 또한 주 모델이 실패할 때 보조 모델이나 규칙 기반 로직으로 전환하는 폴백, 반복 요청에 대한 캐시, 장애 중인 API에 대한 반복 호출을 막는 서킷 브레이커도 함께 다룬다. 마지막으로 응답 시간, 모델별 오류, 재시도 횟수 등을 기록하고 모니터링해야 안정적으로 운영할 수 있다고 강조한다.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Guía paso a paso para instalar Tailscale en un homelab Linux

Este procedimiento explica cómo instalar el cliente de Tailscale en un equipo Linux de un homelab, autenticarlo en la tailnet y verificar su funcionamiento. Como referencia se usa la release v1.104.1, aunque las fuentes no detallan cómo instalar esa versión concreta, por lo que los pasos deben tomarse como el procedimiento general de la documentación. Se requiere un gestor de paquetes compatible, acceso con sudo y el módulo de kernel tun, que puede cargarse manualmente o al arrancar. La instalación se realiza con el script oficial de una línea, copiado de la web de Tailscale, y se autentica con tailscale up, comprobando luego el dispositivo en la consola de administración. Finalmente, la guía recoge opciones como SSH, la gestión de caducidad de claves y las herramientas tailscale ip y tailscale status para revisar el estado.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

d3-turbo: 26B-parameter model answers many questions from one read, outpacing larger rivals on throughput

The vLLM Semantic Router team released d3-turbo on Hugging Face on 19 October 2026, a 26-billion-parameter model with a 460-million-parameter image encoder under an Apache 2.0 license. Rather than generating free-form text, it returns probability distributions over options the user predefines, such as routing a customer ticket or flagging urgency. Its sibling d3-flash re-reads the input for every question, while d3-turbo reads once and answers many, which the author's benchmarks on a single AMD MI325X show becomes faster beyond about four questions. At 255 questions, the reported median latency was 2.12 seconds for d3-turbo versus 32.94 seconds for d3-flash, though d3-flash was quicker for a single question. The model supports choice, yes/no, score, box, and tracking tasks, and accepts up to 131,072 tokens of context when long-context mode is enabled.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

OpenAI Acquires Orchestra Bio to Help Trace Drug Development Delays

OpenAI announced its October 20 acquisition of Orchestra Bio, describing a use case for coordinating drug development. In its example, a failed manufacturing batch could delay a clinical trial, and Orchestra's agents would help identify affected activities and timelines while scientists and program leaders make the decisions. OpenAI also said it plans to combine Orchestra with its Rosalind product to link scientific findings with studies, decisions and next steps. The company has not published a controlled evaluation or quantified time savings for this workflow. The article also offers a proposed developer design for recording impact claims as inspectable, correctable and withdrawable records, though it is not Orchestra's actual architecture.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Phishing and BEC Attacks Bypass Code Defenses by Targeting Employees

Attackers can defeat strong technical security by manipulating staff rather than systems, using impersonation and urgency to trick people into acting. Business Email Compromise is a targeted form of this, where criminals study a company's payment workflows and pose as executives or suppliers to request wire transfers or bank detail changes. The FBI's Internet Crime Complaint Center reported more than $50 billion in exposed losses from BEC between late 2013 and the end of 2022. The article argues that the main defenses are procedural rather than technical, including verifying requests through independently known contact numbers, enabling multi-factor authentication, requiring dual approval for payments, and configuring SPF, DKIM, and DMARC email authentication.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Testing API Failure Scenarios: Why Success Paths Aren't Enough

Developers often verify only the successful 200 OK response, but production systems regularly face server errors, missing data, and network problems. The article illustrates how a simple frontend fetch can crash or leave users stranded when responses are unexpected. It recommends deliberately testing cases such as 500 errors, absent JSON fields, expired sessions, interrupted connections, and requests sent to the wrong environment. Mocking API responses lets teams simulate these conditions without relying on a live backend. Network interception tools such as Proxyceptor are presented as another option for modifying requests and responses without changing application code.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Fan Wiki for Ride A Pet Shows Evidence Labels and Marks Unknown Values

A fan-made guide for the Roblox game Ride A Pet treats every displayed fact as something that must be traceable to a source and date. Each entry is tagged as recorded from gameplay footage, reported by a dated third party, drawn from official artwork, or unknown. Missing measurements appear as "Unknown" rather than zero, so unmeasured stats cannot distort comparisons. Shop and pricing snapshots carry creator, timestamp, and separate "checked" and "last edited" dates to show how current they are. The approach is presented as a data-modeling problem aimed at keeping readers from mistaking guesses for confirmed information.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Study Tests Whether Task Rules Hold When the AI Model Is Swapped

A new paper in the Artificial Deterministic Intelligence series examines whether a bounded operational contract survives when the underlying inference model changes. Task state is kept outside the model weights, and the researchers do not assume smaller models match larger ones in capability. In synthetic first-party tests, a frozen governed package was moved between 8B and 35B models in both directions, with same-model restart controls, and all 120 confirmatory handoff trials kept the measured policy contract and task correctness. The study also compared governed scoped context, authorized metadata-filtered lexical retrieval, and an oracle ceiling, with every configuration completing 90 of 90 trials and conventional lexical retrieval showing lower retrieval-component time. The author states these results do not establish general model equivalence, production readiness, universal safety, or retrieval superiority, and the companion report includes non-acceptances, timeouts, and a non-confirming replication attempt.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Explores Using macOS Touch ID and Secure Enclave for SSH Authentication

A developer has described a method for using Touch ID on macOS to authenticate SSH sessions, with private keys held in the Secure Enclave. The approach aims to avoid the trade-off between security and convenience that comes with passphrases and unprotected keys. The article notes that the Secure Enclave isolates keys from the main operating system, though integration relies on tools such as ssh-agent and custom scripts that could introduce failure points. The author is seeking expert feedback on the method's technical viability and its edge cases. The piece argues that robust yet user-friendly authentication is increasingly important as remote work and cybersecurity threats grow.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Agentic terminals in 2026: seven questions for judging AI coding tools

AI in the terminal has moved beyond command autocomplete to agents that can build an app from a plain description. The article proposes seven questions for telling apart tools that merely suggest code from those that complete the work. Key criteria include whether an agent plans, runs, tests and prepares a project for shipping, and whether a lead agent can coordinate parallel workers. It also highlights multi-model support with your own API keys, the ability to diagnose and fix bugs with configurable approvals, and continuity of work across terminal, IDE, SSH and mobile. The piece is written by the founder of Mel, an agentic terminal tool, who promotes it as meeting these criteria.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Building a Safe Linux Sysadmin Assistant for AI Agents Using a Python MCP Server

AI coding assistants like Claude and Cursor can diagnose errors and suggest system configurations, but giving them raw shell access causes problems such as hung terminals from interactive pagers, excessive token use from oversized log output, and runaway background processes. The article proposes using Model Context Protocol (MCP), an open standard for connecting AI models to local tools, to mediate these interactions. The author's SysFlow MCP server, written in Python, forces non-interactive environment settings on every subprocess to prevent pager freezes. It also enforces hard timeouts that terminate process trees, reads only the tail of large log files with optional regex filtering, and limits service inspection to user-level systemd units rather than root-level daemons.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Quanta Magazine Explores How Hilbert Space Frames Quantum Possibilities

Quanta Magazine has published an article titled 'In Hilbert Space, All Things Are Quantumly Possible' on its website. The piece centers on Hilbert space, the mathematical framework that quantum physics uses to describe the states a system can occupy. The only details available in the source material are the article's URL, a Hacker News discussion thread, and its current point and comment counts. The full article text was not included, so the specific arguments and findings it makes cannot be summarized here.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

Claude Code subagents, hooks and slash commands: a practical setup guide

The guide explains how to turn CLAUDE.md guidance into enforced behavior using subagents, hooks, and slash commands in Claude Code. Subagents are Markdown files in .claude/agents/ with YAML frontmatter, each running in its own context with limited tools such as read-only access for reviewers. Hooks configured in .claude/settings.json can block or allow tool calls, such as stopping destructive shell commands or refusing edits to secret files, and can run formatters or tests. Slash commands defined in .claude/commands/ chain subagents into repeatable workflows. The article also advises reading hook scripts before enabling them and keeping them under version control, since hooks run with the user's permissions.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AGENTS.md, CLAUDE.md, or .cursor/rules: Which AI Agent Reads Which File?

Different AI coding tools look for different instruction files, so keeping rules in just one tool's format leaves the others guessing. Copying identical text into every format, however, tends to cause the versions to drift apart over time. AGENTS.md is an open, cross-tool Markdown convention read by OpenAI Codex, Cursor, and other agents. Claude Code uses CLAUDE.md, which can import other files such as AGENTS.md, while Cursor uses scoped .mdc rule files with YAML frontmatter, and Windsurf stores its rules as Markdown under .windsurf/rules/. The article recommends a maintainable setup that keeps shared guidance in AGENTS.md and layers tool-specific notes on top.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Java Encapsulation and Visibility: Protecting Object State with Access Modifiers

The article explains how encapsulation and visibility in Java help developers control how objects are used, not just whether code works. It uses a BankAccount class as an example, where the balance field is private so it cannot be changed directly from outside the class. Methods are given different access levels: public for deposits and withdrawals, package-private for applying fees within the same package, and protected for subclasses and same-package classes. A private method, resetBalance, is shown as an internal implementation detail hidden from other code. Together, these modifiers protect the object's data and limit how the class can be used.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

LocalRAG Desktop Beta Offers Docker-Free Private Document Q&A for Windows

A developer has released LocalRAG Desktop, a beta local document question-answering app for Windows that runs without Docker, WSL, or Python. Users can upload text-based PDF, Word, TXT, or Markdown files, which are parsed, chunked, and embedded locally before being searched via Qdrant. Answers are generated by a local Qwen model, with the source file and chunk shown for each response. All services bind to 127.0.0.1, use a session token for API access, and send no default telemetry. The beta supports Windows 10 22H2 and Windows 11 x64 with 8GB RAM recommended, and is not yet code signed, so Windows may display an unknown publisher warning.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Kicking Participants From Video Rooms Needs Server-Side Authorization, Not Client Notices

The article argues that removing a participant from a video room should revoke server-side authorization rather than just send the browser a leave message. It recommends issuing narrowly scoped admission tokens and tracking a membership epoch on the server, so reconnections carrying an outdated epoch are rejected. A tabletop example shows a kicked contractor rejoining within seconds using a still-valid credential, because transport closure was mistaken for authorization revocation. The author stresses that clients cannot be trusted to enforce access and that alerts should detect revoked memberships that regain access. Idempotent moderation commands are also recommended so retries do not produce inconsistent outcomes.

0 comments Read more at DEV Community
← NewerPage 206 of 2991Older →