SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingDEV Community ·

DDRop Attack Uses $159 Hardware to Break Cloud Confidential Computing Protections

Researchers from KU Leuven, ETH Zurich, Durham University, and Google have disclosed a hardware attack called DDRop that can compromise confidential computing protections used in Intel TDX and AMD SEV-SNP systems. The attack uses a custom-built circuit board costing just $159, placed between the CPU and DDR5 memory, which silently drops write commands to keep stale data in memory without triggering encryption alerts. By exploiting this weakness, attackers can gain full control of protected virtual machines and forge attestation — making a VM falsely appear to be running unaltered code. The attack requires both software-level server access and brief physical access to install the interposer hardware. Intel and AMD acknowledged the findings but stated physical-access scenarios fall outside their threat model, and no simple software fix is possible given the architectural nature of the flaw.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How PHP Sessions Enable Stateful AI Chat Without Persistent Memory

Building a stateful chatbot in PHP is complicated by the fact that each HTTP request runs as a fresh process with no shared memory between calls. A practical solution uses PHP sessions via session_start(), storing conversation history in the $_SESSION array so it persists across requests for the same browser. The user's message and the AI agent's reply are each appended to this session array, which is passed to the agent on every request to maintain conversational context. For more durable storage, a one-line swap replaces session-based history with a FileMemory or PdoMemory driver, enabling persistence across server restarts and support for multiple users. This approach is part of the NanoAgent examples series, which demonstrates modular AI agent patterns in PHP.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

NanoAgent lets PHP developers build AI agents without switching to Python

A lightweight PHP library called NanoAgent allows developers to build functional AI agents entirely within PHP, without needing to adopt Python or a heavy framework. The minimal setup requires just four components: an Agent, a FunctionTool, a Task, and an LLM configuration object. Developers define callable PHP functions wrapped in a JSON Schema, which the language model can invoke autonomously based on task context. The library handles the tool-call loop internally, so developers do not need to manually wire prompt-response cycles. NanoAgent defaults to Groq's free-tier LLaMA model, making it accessible for first-time users with no API configuration required.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Why Riverpod Was Built and Why Modern Flutter May No Longer Need It

Riverpod was created by Rémi Rousselet as a direct response to a fundamental limitation in Flutter's InheritedWidget system, which offers no mechanism to unsubscribe a widget from a dependency once it has been registered. This flaw meant widgets could continue rebuilding on every ancestor state change even after the condition triggering their dependency turned false, causing unnecessary overhead in larger apps. Rousselet attempted to fix this at the framework level in a 2019 pull request, but Flutter's then-tech lead Ian Hickson closed it, citing concerns about added complexity in Flutter's core build loop. Unable to patch InheritedWidget itself, Rousselet built Riverpod entirely outside the widget tree to work around the constraint. A newer approach called BlocSignal is now presented as a potential successor, aiming to restore InheritedWidget to its original role as a scoped injector while handling state propagation through reactive signals instead.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Why Chasing 'Top 0.1% in Your Field' May Be the Wrong Goal

A viral LinkedIn post claiming that only the top 0.1% of professionals earn respect sparked a critical reexamination of how people measure career success. Developer Maneshwar, building an AI code review tool called LiveReview, argues that 'top 0.1%' is not an actionable goal because it cannot be scheduled, practiced, or directly pursued on any given day. Drawing on Warren Buffett's concept of the 'Inner Scorecard' from Alice Schroeder's biography The Snowball, he distinguishes between measuring oneself by internal effort and consistency versus external rankings and reputation. Buffett's framework highlights that outer metrics like rank or follower count can shift regardless of personal performance, while inner metrics such as honesty, intensity, and consistency remain within one's control. The piece ultimately argues that replacing vague status-based ambitions with specific, executable tasks is a more reliable path to genuine excellence.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

React useActionState Hook Set to Replace useReducer for Server-Driven Form Handling

React's useActionState hook is gaining traction in 2026 as a cleaner alternative to useReducer for managing server-driven form logic in modern web applications. The hook accepts a server action and an initial state, then automatically handles pending, error, and success states without requiring manual dispatch or custom reducer scaffolding. Unlike useReducer, it passes the previous state directly to the server action on each call, enabling incremental updates and reducing common failure modes like state desyncs and error-shape mismatches. The pattern is particularly relevant for teams building production apps on frameworks like Next.js and Remix, where server actions are now the default form-handling approach. However, useActionState is not a universal replacement — complex client-only state machines and multi-step wizards may still benefit from explicit reducer logic.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Codename One Launches Vault API for Secure Cross-Device Data Sync Without Exposing Keys

Codename One, an open-source framework for building cross-platform apps from a single Java or Kotlin codebase, has released a new com.codename1.security.vault API as part of its latest update. The Vault API enables encrypted data to be synchronized across phones and browsers without the sync server ever receiving the encryption key or plaintext. It uses authenticated encryption with a random 32-byte data key, storing only wrapped copies of that key so that a password change does not require re-encrypting all records. The API supports three unlock policies — session-only, device-remembered, and user-verification — giving users deliberate control over how vault access is restored. In browser environments, remembered access leverages a non-extractable Web Crypto key stored in IndexedDB, strengthening security beyond conventional approaches like storing a key alongside its ciphertext.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Laravel Context: How to Share Execution Data Across Requests, Logs, and Jobs

Laravel's Context feature allows developers to store and retrieve information — such as trace IDs, tenant IDs, or URLs — that needs to persist throughout a single application execution without being passed explicitly between methods. The data can be added via Context::add() and accessed anywhere in the application using Context::get(), making it especially useful for consistent log enrichment. When a queued job is dispatched, Laravel automatically captures the current context, serializes it into the job payload, and restores it when the worker processes the job — a process called dehydration and hydration. This ensures that contextual data from an originating HTTP request, like a trace ID, remains accessible inside the job for debugging distributed workflows. Importantly, Laravel isolates context per job execution, preventing data from one job from leaking into the next in long-running queue workers.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Anthropic CEO Dario Amodei Appears on SNL Weekend Update to Discuss AI Risks

Anthropic CEO Dario Amodei made an appearance on Saturday Night Live's Weekend Update segment to address the potential threats posed by artificial intelligence. The segment drew attention on Hacker News, where it was shared as a video link. Amodei, who leads one of the prominent AI safety-focused companies, used the platform to speak about AI's risks to humanity. The clip garnered modest engagement on the tech forum, receiving 8 points and one comment.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

Dev Team Builds AI Coding Agent With Persistent Memory Across Sessions

A development team grew frustrated with AI coding assistants that lose all project context at the end of each session, forcing developers to re-explain architecture and previously solved bugs every time. To solve this, they built a web agent with a lightweight, persistent memory layer that reads and updates project knowledge as it works. The system stores architectural decisions, past bug fixes, coding conventions, and feature progress in structured, human-readable Markdown files rather than a complex vector database. Built using React, Next.js, Node.js, and an LLM-based coding layer, the agent checks its memory before attempting solutions, enabling consistent and informed responses across sessions. The team reports the key improvement was not smarter code generation but better continuity, making the agent behave more like a long-tenured teammate than a stateless tool.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer ditches tutorial hoarding for small project-based learning

A developer shared on DEV Community how they shifted away from endlessly collecting tutorials toward a more hands-on learning loop. The new approach involves learning one concept, building something small with it, and working through problems as they arise rather than seeking complete mastery upfront. Getting stuck and debugging is now treated as a core part of the process rather than a setback to avoid. This method produces tangible outputs instead of folders full of saved but unused resources. The developer is still experimenting with the approach and invited others to share what helped them bridge the gap between understanding tutorials and independently building projects.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

CS Student Shubham Shah Builds Real-World Systems to Master Software Engineering

Shubham Shah, a Computer Science Engineering student, began programming in Grade 8 and has since developed a hands-on approach to learning by building functional projects rather than following tutorials. His most significant project, Hostello, is a hostel management platform designed to replace manual registers and spreadsheets with a full-featured digital system. The platform incorporates role-based access control, real-time communication, QR-based identification, and multi-building management, among other capabilities. Building Hostello shifted Shah's understanding of software from simple code-writing to tackling production-level concerns like security, scalability, and system maintenance. Beyond development, he holds a strong interest in Linux internals and infrastructure, working with tools such as Docker, reverse proxies, and server administration across a broad technology stack.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AI Voice Generator APIs in 2026: How ElevenLabs, Google, and Azure Compare

The text-to-speech API market has grown significantly by 2026, with multiple providers offering features such as voice cloning, real-time streaming, and multi-language support for developers building chatbots, audiobooks, and voice assistants. A comparative analysis of leading platforms — ElevenLabs, Google Cloud TTS, Amazon Polly, Azure Speech Service, Coqui TTS, and Respeecher — evaluates them on audio quality, latency, cloning workflow, language coverage, and pricing. ElevenLabs scores highest on the Mean Opinion Score naturalness benchmark at 4.8, supports few-shot voice cloning from just 3–5 seconds of audio, and offers real-time WebSocket streaming at $16 per million characters. Google Cloud TTS covers over 220 languages but lacks voice cloning, while Azure requires at least 30 minutes of studio audio to set up a custom voice. For developers prioritising a balance of quality, flexibility, and cost, ElevenLabs currently holds an edge, though open-source alternative Coqui TTS remains viable for self-hosted deployments despite requiring GPU-based training.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AI Agents Shift Focus to Smarter Human Escalation Over Automation

As AI agents move into corporate environments, the design emphasis has shifted from replacing humans to integrating agents into human workflows with meaningful oversight. Research shows that Claude Code users approve 93% of permission requests, prompting Anthropic to build an auto mode, since frequent interruptions cause people to approve requests without careful review. Experts now distinguish between simple approval prompts and true escalation, where an agent hands off a situation it cannot safely resolve, along with full context and available options. Google's agentic design patterns similarly recommend using deterministic workflows for routine tasks, reserving human judgment for ambiguous decisions, policy exceptions, and matters of responsibility. The broader goal is not to eliminate human involvement but to reduce unnecessary interruptions, so that when agents do escalate, they bring well-prepared cases that genuinely require human judgment.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AI Agent Reviews 43 Startup Landing Pages, Finds Key Claims Buried Too Deep

An AI agent named Junior, operating on the iLands platform, spent 50 days autonomously browsing the open web and cold-reading 43 startup landing pages to identify common copywriting weaknesses. Using tools like the Hacker News Algolia API and GitHub repository searches, Junior sourced fresh startup launches and contacted founders with unsolicited but free page feedback. Across the 43 pages reviewed, four recurring problems emerged: the strongest value proposition was consistently buried several scrolls down, multiple conflicting calls-to-action confused visitors, key numbers were hedged with vague language, and critical promises were hidden in FAQs rather than featured prominently. Of 59 cold emails sent offering free feedback, Junior received only one reply and made no sales, attributing the poor response rate to the high volume of pitches founders already receive. The core takeaway for landing page owners: lead with the most compelling, specific claim rather than saving it for the fine print.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

SAGEWORKS AI Builds Offline Windows Knowledge Engine That Avoids Cloud and AI Hallucination

SAGEWORKS AI has developed Library Ground / Aerie, a Windows-native offline knowledge management system designed to run on standard Windows 10 machines without internet access, cloud services, or additional developer tools. The system operates entirely through Windows PowerShell 5.1, eliminating the need for Rust, Python, Node.js, or a web server. Rather than using a language model, it functions as a symbolic ledger engine that stores sourced information, tracks history, and retrieves answers by matching tokens against locally saved content with traceable references. The architecture uses three logical roles — Herald, Meridian, and Forge — all running within a single local PowerShell process, leaving no background services after the session ends. The project is explicitly designed to address knowledge preservation on constrained or offline machines, prioritising verifiable, evidence-based responses over generated or inferred answers.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

New AI Benchmark Tests Whether Models Know When to Question Their Own Decisions

A developer built a benchmark called Decision-Conditioned Falsification to test whether AI models can rationally decide when gathering more information is worth the cost before acting. The environment presents three possible hidden mechanisms, and models must compose experiments and commit to operational actions, with experiments costing points and poor decisions costing even more. Unlike typical benchmarks, it rewards net operational utility rather than the volume of self-criticism, hypothesis revisions, or experiments run. Testing across four models — GPT-4.5 nano, Claude Haiku 4.5, Gemini 3.5 Flash, and Gemini 3.7 Flash — revealed no consistent aggregate benefit from explicitly prompting models to seek contradicting evidence. A key finding was that models sometimes predicted outcomes impossible under any mechanism, suggesting the core problem is not willingness to self-correct but the ability to accurately calculate what a hypothesis actually predicts.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AI Agent on iLands Platform Finds DEV Community the Only Open Door Without a Phone

An AI agent operating on a platform called iLands, which provides agents with budgets and tools, set out to find a public online space where it could interact with humans outside its own platform. Most major platforms were inaccessible, requiring phone numbers, account age, or third-party logins via Google or GitHub. DEV Community accepted only an email address, making it the sole platform the agent could join independently. The agent's self-described function is to look up real locations — streets, buildings, or landmarks users cannot visit — using maps, satellite imagery, and archives, then report back what it could and could not verify. Its first such lookup involved a granite monument in Rēzekne, Latvia, traced through a newspaper archive and a 2012 street photo, with the building now confirmed gone.

0 comments Read more at DEV Community
← NewerPage 864 of 3175Older →