SShortSingh.

Programming

0
ProgrammingDEV Community ·

Why AI Agents Need an Open Task Market to Coordinate Work and Payments

As AI agents grow more capable, they are expected to move beyond isolated chat interfaces and begin hiring other agents or people to complete tasks they cannot handle alone. This vision, described as an 'agentic economy,' relies on open task markets where work is posted with clear acceptance criteria, funded rewards, and verifiable results. To address the classic trust problem between buyers and workers, such markets would use smart contracts to hold funds and enforce settlement rules transparently, rather than relying on slow human arbitration. Early viable use cases would likely focus on deterministic, machine-checkable work such as code patches, data formatting, or file verification. Tasks requiring creative or subjective judgment would still need human review, meaning hybrid systems rather than full automation would define the near-term reality.

0
ProgrammingDEV Community ·

Build a 60-Line Python Switchboard to Route AI Tasks Across LLM Providers

A developer has shared a minimal, standard-library-only Python pattern for routing AI tasks to different large language model backends without rewriting test scripts each time a new model launches. The switchboard, around 60 lines of code, maps task types such as summarize or code to a prioritized list of backends, automatically falling back to the next option if one fails. No third-party packages or API keys are required to run the core experiment, making it accessible for quick evaluation. The approach separates what is being asked from which model answers it, so evaluation questions remain stable even as backends change. The tutorial includes a failing fallback fixture to demonstrate exactly where and why routing breaks under edge cases like empty input.

0
ProgrammingDEV Community ·

Recraft-AI Launches Clarity-Upscale Model for Sharpening Images on Replicate

Recraft-AI has released an image upscaling model called Recraft-Clarity-Upscale, available on Replicate's cloud infrastructure. The model takes a hosted image URL as input and returns a sharpened, cleaner version optimized for web publishing or print production. Unlike more advanced upscalers, it focuses strictly on edge clarity and sharpness rather than adding new textures or reconstructing missing detail. It is suited for use cases such as product photography, print-ready asset preparation, screenshot enhancement, and batch image refinement. The model offers no adjustable parameters, and output quality depends significantly on the clarity of the original source image.

0
ProgrammingDEV Community ·

StayPresent v1.6.0 Adds Crash Log Viewer to Status Page, No SSH Needed

StayPresent v1.6.0 introduces a built-in crash log feature that captures bot stdout and stderr into a ring buffer, making recent output accessible directly from the platform's status page. Previously, bot output was passed straight through to the parent console, meaning logs could scroll past and be lost before a developer noticed a crash. The new ring buffer retains recent log lines per bot while keeping memory usage bounded, regardless of how long the bot has been running. When a crash occurs, the status page's admin view now displays a tail of recent output alongside the exit code, eliminating the need for shell access to diagnose the failure. Developers should note that bots now always receive False from sys.stdout.isatty(), since piped streams are required to enable the capture functionality.

0
ProgrammingDEV Community ·

Recraft-AI Launches Crisp-Upscale Model for Sharp, High-Res Image Enhancement

Recraft-AI has released an image upscaling model called Recraft-Crisp-Upscale, hosted on Replicate's cloud infrastructure and accessible via API without requiring local GPU hardware. The model takes a single image as input and reconstructs detail to produce higher-resolution output, rather than simply stretching existing pixels. It is designed for professional use cases such as web asset preparation, print-ready material production, logo enhancement, and restoration of legacy low-resolution images. The model operates with a fixed configuration, offering no parameters to adjust upscaling intensity or output resolution, and cannot process video, batch images, or invent content absent from the source. For use cases requiring creative detail generation beyond the original image, Recraft-AI recommends its separate Recraft-Creative-Upscale model instead.

0
ProgrammingDEV Community ·

A Bay Area Engineer Reflects on 15+ Hackathons and What They Actually Teach

A software engineer at Cisco, who previously attended SJSU and interned at Volvo Cars, has competed in more than 15 Bay Area hackathons since 2018, winning four of them across events hosted by Google, Microsoft, Oracle, and others. Writing for DEV Community, he describes the typical arc of a hackathon weekend: ambitious plans on Saturday, heavy cuts by afternoon, and the real grind beginning around midnight. One memorable all-nighter ended with 40 minutes spent tracking down a single misconfigured value set differently across two files, only to miss a placement after demoing at 10am. He argues the real value lies not in prizes or projects, but in the collaborative intensity of building alongside strangers under pressure and no hierarchy. According to him, the experience sharpens practical skills like rapid decision-making and ruthless feature-cutting in ways that conventional work environments rarely replicate.

0
ProgrammingDEV Community ·

Agent Memory Leaderboard Launches to Standardize AI Memory System Evaluation

A new benchmarking platform called Agent Memory Leaderboard (AML) has been launched to enable fair, consistent comparisons of AI agent memory systems. Jointly initiated by nearly 30 universities and research institutions, AML separates memory retrieval from answer generation and evaluation to eliminate confounding variables. The first evaluation round saw 67 memory frameworks complete testing across open-source and commercial tracks, with 136 teams registered in total. AML assesses memory systems across multiple dimensions including factual recall, temporal reasoning, personalization, and epistemic safety — not just retrieval similarity. The first leaderboard results went live on August 12, 2026, attracting over 100,000 clicks within the first ten days.

0
ProgrammingDEV Community ·

Structured Data Models Can Make Survival Game Item Guides Far More Useful

A developer has outlined a data-driven approach to building item guides for survival games like Subnautica 2, arguing that key facts such as unlock conditions, required equipment, and navigation routes should be stored as structured fields rather than buried in prose. The method separates prerequisites, landmarks, hazards, and recovery steps into distinct data categories, making guides easier to follow and errors easier to diagnose. Using the Repair Tool as a case study, the author demonstrated how navigation can be broken into discrete segments, each with its own origin point, bearing, and visual confirmation landmark. The approach also distinguishes between mandatory and recommended equipment, preventing beginners from delaying progress due to unclear requirements. Hazard warnings, the author argues, are most useful when attached directly to the relevant route segment rather than grouped into a generic notice.

0
ProgrammingDEV Community ·

Developer builds browser-style DevTools dashboard for Node.js backend debugging

A developer frustrated with console.log-based backend debugging has released Wevna, an open-source local observability tool for Node.js applications. The SDK requires just two lines of code to set up and opens a localhost dashboard displaying HTTP requests, SQL queries, Redis commands, and logs in a visual, waterfall-style interface. Wevna automatically flags patterns such as N+1 database queries by detecting repeated queries with the same shape within a single request. The tool is privacy-focused by design, with all data staying on the local machine and no account, API key, or external data transfer required. Currently in v1, it supports Postgres and Redis only, and the developer is actively seeking feedback from users testing it on real applications.

0
ProgrammingDEV Community ·

Developer Builds C# .NET App to Help Shoppers Compare Prices and Travel Costs

A developer has created a web application called Digital Fare App using the C# .NET ecosystem, designed to simplify everyday purchase calculations. The tool helps users determine whether bulk deals or longer trips with multiple passengers are genuinely cost-effective. Users can save their calculation records to a database for future reference. Auth0 integration for authorization and user authentication is planned as an upcoming feature. The project is currently seeking community input on branding, including a name and logo.

0
ProgrammingDEV Community ·

How to Build a Lightweight AI Agent in 150 Lines of Python Without Frameworks

A software developer shares a step-by-step guide to building a production-ready AI agent in approximately 150 lines of Python, using no external frameworks such as LangChain or CrewAI. The approach was inspired by a startup founder who struggled with framework abstractions that did not fit his specific workflow. The hand-built agent includes tool use, working memory, budget controls, and escalation logic, with every line remaining fully readable and traceable. Building from scratch gives developers direct visibility into model context, honest budget enforcement, and easier debugging compared to opaque framework internals. The author acknowledges that frameworks can save time for standard workflows, but argues that writing the core loop manually — roughly 50 lines — yields greater control and understanding.

0
ProgrammingDEV Community ·

Fake USB and RDP Devices Can Grant Attackers SYSTEM Privileges on Windows

Security researchers have demonstrated a technique called 'Plug and Pwn' that exploits Windows Plug and Play to gain SYSTEM-level privileges by connecting fake or virtual USB devices. The attack works by tricking Windows into recognizing spoofed hardware — such as a Sierra Wireless or Sony FeliCa device — prompting automatic signed driver downloads from Windows Update and installation with SYSTEM privileges. A second attack chain abuses RDP USB redirection to deliver a crafted virtual device descriptor, exploiting Intel RealSense driver co-installers to execute malicious code without physical access. Neither user login nor UAC approval is required, making the attack appear as routine device setup behavior to the operating system. Mitigations include disabling RDP PnP redirection, enforcing device installation restrictions by Hardware ID, blocking Windows Update driver downloads, and applying vulnerable driver blocklists.

0
ProgrammingDEV Community ·

City-Forum Campaign Exploits Salesforce and ServiceNow Guest APIs to Harvest Data

A threat actor tracked as City-Forum is using a custom Go-based tool to silently enumerate publicly accessible data on Salesforce Experience Cloud and ServiceNow Service Portal instances without requiring authentication. The tool exploits misconfigured guest user permissions in Salesforce Aura and LWR GraphQL APIs to extract records including accounts, contacts, cases, leads, and documents left visible to unauthenticated users. It simultaneously probes ServiceNow's public Service Portal search API to collect data exposed through Knowledge Bases and custom search sources. All traffic originates from IP address 158.220.87.79 hosted on Contabo, linked to the domain city-forum.com since March 2025, with no malware deployed on victim systems and no user login events generated. Security researchers recommend disabling LWR guest UI APIs, enforcing minimum privilege for guest users, and adding login gates to ServiceNow search sources to mitigate the risk.

0
ProgrammingDEV Community ·

Android Malware Duo SpyNote and WindRelay Used to Steal Loans and Relay Card Payments

Security researchers at Group-IB have uncovered a sophisticated Android fraud campaign combining two malware tools, SpyNote and WindRelay, to commit financial theft. Attackers impersonate bank staff over phone calls and trick victims into sideloading a personalised SpyNote APK and granting Accessibility Service permissions. Once installed, SpyNote silently deploys WindRelay and remotely operates the victim's official banking app to take out loans in their name. WindRelay captures NFC card data in real time when the victim taps their physical card on their phone, relaying it to the attacker's device, which emulates the card at a legitimate point-of-sale terminal. The attack requires no root access and exploits standard Android features, making it difficult for banks to distinguish fraudulent activity from normal device behaviour.

0
ProgrammingDEV Community ·

Lazarus Group Uses Windows Zero-Day and Fake Job Offers to Backdoor Defense Firms

North Korea-linked Lazarus Group has been running 'Operation Dream Job', targeting defense and aerospace personnel with fake job offer PDFs to deliver a multi-stage malware chain, according to Check Point Research published on August 11, 2026. Victims are tricked into running a signed PDF viewer that side-loads a malicious DLL, which silently executes the MISTPEN backdoor in memory while displaying a decoy document. MISTPEN communicates with attacker-controlled servers via Microsoft Graph API and OneDrive, before exploiting a critical Windows 11 kernel zero-day, CVE-2026-68820, to gain SYSTEM privileges and disable endpoint detection tools using the FudModule rootkit. The final payload, a backdoor called ForestTiger, is then deployed alongside the Troy backdoor, which supports 17 command types including file transfer, process reconnaissance, and in-memory DLL injection. Attackers also compromised Roundcube mail servers by exploiting CVE-2025-49113 to install the RelayShell web shell, using them as traffic relays to obscure command-and-control infrastructure.

0
ProgrammingDEV Community ·

Developer Builds Repeatable Test Harness to Objectively Compare Free Coding AI Models

A software developer found that standard demo prompts gave misleadingly positive impressions of AI coding models, which often failed on real-world tasks. To address this, he built a lightweight Python harness that runs a fixed set of realistic tasks against multiple models and logs the results for comparison. The task suite includes constraint-following, style-matching, bug localization, and a deliberate impossible prompt designed to expose hallucination under ambiguity. Keeping both the tasks and prompt templates frozen ensures that only the model variable changes between evaluations, making comparisons meaningful. The approach is designed to help developers decide whether a free-tier model is genuinely suitable for their specific workflow before integrating it.

0
ProgrammingDEV Community ·

Developer builds circuit breaker to stop free AI models failing silently in CI pipelines

A developer discovered that a free AI model integrated into a CI pipeline was silently producing empty changelog entries for a week, with the pipeline reporting no errors. Unlike paid APIs with SLAs, free model endpoints can degrade without triggering alerts, causing automation to ship broken or empty outputs undetected. To address this, the developer built a Python-based router featuring a canary quality gate and a circuit breaker that probes the model before each real request and halts calls after repeated failures. The system routes to a safe deterministic fallback rather than blocking the build when an endpoint is deemed degraded. The approach targets low-stakes CI tasks such as changelog generation and commit message linting, where silent failure is more damaging than occasional mediocre output.

0
ProgrammingDEV Community ·

CS Student Contributes to Open Source Projects Using Only a Smartphone

Melwin Siril Joel, a second-year Computer Science Engineering student at Annamalai University in Tamil Nadu, India, has been actively contributing to open source projects using only an Android phone. Unable to always access a laptop, he built a mobile-based workflow using Termux and GitHub Mobile to clone repositories, write code, run tests, and submit pull requests. He has had pull requests merged in real projects, including MapLibre Martin, and participated in programs such as FOSShack 2026 and GSSoC 2026. Beyond coding, Joel serves as an NCC Army Wing Cadet and represents his university in competitive chess. He is also exploring initiatives to bring open source culture to his campus through communities like FOSS United Campus Clubs and GDG on Campus.

0
ProgrammingDEV Community ·

Developer Builds Low-Power Android Geofencing Engine to Auto-Manage Sound Profiles

A developer has detailed the creation of 'Muffle', an Android app that automatically adjusts sound profiles based on the user's physical location. The project was motivated by the cognitive burden of manually managing device notifications across different contexts such as meetings, prayer, and study sessions. Rather than using battery-draining GPS polling, the app leverages Android's GeofencingClient from Google Play Services, which offloads proximity detection to hardware-level system services. The app uses a PendingIntent and BroadcastReceiver pattern, allowing it to remain dormant until a geofence transition is detected, then triggering the AudioManager to adjust volume. A minimum geofence radius of 100 meters was set to prevent erratic toggling caused by GPS drift, though OEM inconsistencies across Android devices posed additional development challenges.

0
ProgrammingDEV Community ·

Clay Board Style System v0.4.1 Brings CSS-Like Animations to Native SDL3 UI

Clay Board Style System (CBSS) version 0.4.1 was released on August 12, 2026, introducing a new SDL3 demo that demonstrates CSS-inspired keyframe animations and hover transitions running inside a native UI built with Nim. The release builds on v0.4.0, which landed a day earlier and introduced the core declarative motion runtime. The demo showcases four animation behaviors in a single native SDL3 window, including eased horizontal movement, 2D flips, color and opacity transitions, and reversible hover effects. CBSS keeps motion declarations within the style model and samples only active animation tracks at their scheduled frame deadlines, avoiding a full style-and-layout pipeline pass for purely visual changes. Version 0.5 is planned to expose the declarative motion system through a C ABI, broadening its accessibility beyond Nim.

← NewerPage 173 of 1337Older →