SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingDEV Community ·

Solo Developer Measures Real AI Coding Gains: 2.4x to 3.2x Productivity Boost

A developer running MachDuDas, a decade-old German marketplace, conducted an informal experiment to measure how much AI coding agents actually improved his engineering output compared to earlier conventional development on the same codebase. Rather than relying on generic industry benchmarks, he compared current AI-assisted commits against historical team contributions from the same product's repository. Between March and September 2026, one human directing AI agents produced roughly 81,000 runtime-code additions and over 105,000 lines of automated tests — far exceeding the test coverage seen in earlier periods. When normalized per active human contributor-month against a productive three-person team from July 2016, the AI-assisted workflow delivered approximately 2.4x to 3.2x the runtime-code output. The author notes that AI also made testing dramatically cheaper to produce, meaning the productivity gains extend beyond raw code volume to include significantly more automated verification.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

HourTotal stores time card data entirely in the URL, requiring no login or backend

A developer built HourTotal, a time card calculator that encodes all work-hour data directly into the page URL as a base64url-encoded JSON fragment, eliminating the need for accounts or server storage. The tool supports up to two punch pairs per day across a five-day workweek, along with pay rate and overtime settings, compressing to roughly 390–650 characters of JSON. Because browsers exclude the URL fragment from HTTP requests, sensitive payroll data never reaches servers, CDN logs, or third-party platforms like Slack. After early issues with Safari's replaceState rate limit, the developer added a 500ms debounce so the URL updates shortly after each edit rather than on every keystroke. Sharing a time card is as simple as sending the link, which reopens the identical filled-in calculator for anyone who receives it.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Polymarket's endDate Field Does Not Mark Resolution Deadline, Analysis Finds

Developer research firm OrcaLayer examined 23 Polymarket markets via the Gamma API on 23–24 September 2026 and found that the endDate field does not represent the resolution deadline as Polymarket's documentation implies. Across sports markets, endDate corresponded to kickoff time for NFL and soccer, scheduled start plus six hours for esports, and original start plus seven days for tennis — none of which matched the actual resolution rules. The true binding deadline for each market is written in plain English within the description field, while trading status is reflected in the closed and acceptingOrders fields. Developers building tools on Polymarket data — including trading bots, dashboards, or predictive models — risk errors if they rely on endDate as a resolution timestamp. OrcaLayer cautioned that endDate's meaning varies by market type and the API itself does not formally document what the field represents.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Scaling AI Agents to a Million Requires Distributed Systems Thinking

A technical blog post published on Instacloud argues that deploying AI agents at massive scale is fundamentally a distributed systems challenge. The article draws parallels between managing large numbers of autonomous agents and classic problems in distributed computing. It suggests that engineering teams building multi-agent systems must apply lessons from distributed architecture to ensure reliability and coordination. The post sparked discussion on Hacker News, where it received early engagement from the developer community.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

Developers Build AI Fraud Investigation Agent Using TigerGraph, MCP, and GraphRAG

A team participating in the TigerGraph × Hacker House Goa 2026 Challenge has built an agentic fraud investigation system designed to go beyond standard anomaly-score detection used in banking. The system combines TigerGraph Savanna Cloud, which stores over 590,000 transactions and 5,500 historical fraud cases, with a large language model agent that reasons iteratively over graph-based evidence. A Model Context Protocol (MCP) layer connects the AI agent to the graph database, enabling multi-hop relationship queries across transactions, cardholders, devices, and past cases. A deterministic policy engine enforces strict bank compliance rules and determines investigation verdicts, including card blocking and regulatory reporting under FinCEN SAR requirements. The architecture deliberately separates probabilistic AI reasoning from authoritative rule-based decision-making to ensure compliance and operational reliability.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Whiteboard (YC W26) launches open-source IDE for human-AI software architecture

A four-person team of former tech leads has released Whiteboard, an open-source desktop IDE designed to help developers and AI agents collaboratively design and review software. Built on top of CodeOSS, the app lets AI agents draw diagrams and traces on an in-canvas workspace, with direct links to the underlying code. Key features include a Rust-based semantic diff viewer, a decision log for tracking autonomous agent choices, and compatibility with tools like Claude Code and Codex. The app is already being used by engineers at companies such as Salesforce and Modal for reviewing architecture and spec-level changes. Released under an MIT license, the team plans to eventually offer a hosted web version with multiplayer reviews and trajectory storage while keeping the tool self-hostable.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

How to Safely Deliver Buyer Images While Preserving Unmodified Originals

A software architecture guide outlines best practices for delivering purchased images to buyers while keeping original files immutable and privately stored. The approach recommends serving only moderated, processed renditions by default, releasing originals solely when a buyer's order explicitly grants that entitlement. Every delivered image must be traceable to a specific source revision, crop specification, and moderation decision to prevent unreviewed content from reaching paying customers. Alerting systems should trigger on confirmed policy bypasses rather than every failed crop, and dashboards must identify failures by ratio and transform version to avoid masking localized errors. The guide also stresses that authorization decisions must precede byte-range handling, and that changing source files or crop algorithms silently invalidates prior moderation coverage.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

TypeSafe AI Launches Jev, a Model Built for Structured Decisions Over Text Generation

TypeSafe AI has released Jev, its first public 'System One' model, designed to make fast, structured decisions that software can consume directly rather than generating natural language responses. The model is inspired by Daniel Kahneman's System 1 and System 2 thinking framework, positioning Jev as a quick, intuitive decision-maker suited for backend applications. Instead of parsing generated text into booleans or scores, developers define typed questions and receive structured outputs such as yes/no answers, categorical choices, or scored evaluations in parallel. Jev uses a new architecture combining a parallel sampler and a training method called Reinforcement Learning for Calibrated Decisions (RLCD) to deliver calibrated probabilities alongside each decision. The model targets use cases like routing support tickets, flagging sensitive prompts, approving agent actions, and triaging transactions where structured outputs matter more than human-readable explanations.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Tutorial: Build an Adaptive Python Tutoring API Using FastAPI and SQLite

A developer tutorial published on DEV Community walks readers through building PyMentor, a lightweight adaptive tutoring API for Python learners. The API, built with FastAPI and SQLite, accepts a learner's code submission along with their identifier and topic, then queries an OpenAI model for structured teaching feedback. It tracks topic mastery scores in a local SQLite database and returns validated responses to the client. The project is intentionally scoped — it does not execute submitted code or replace human instructors, focusing solely on repeatable feedback workflows. Developers need Python 3.10 or later, an OpenAI API key, and basic familiarity with Python, JSON, and HTTP requests to follow along.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Tutorial: Enforce Beauty AR Device Policies in CI Before They Reach Users

A new developer tutorial demonstrates how to build a policy gate for Tencent RTC Beauty AR integrations that prevents expensive effects like GAN and 3D segmentation from running on low-end devices. The approach encodes rollout decisions — such as which effects are permitted per device tier and what fallback profiles apply — directly as versioned, testable code. A lightweight Node.js project structure validates profile configurations, detects cyclic or incomplete fallback chains, and exposes runtime states including applying, degraded, and blocked. The validation runs entirely within GitHub Actions CI without requiring a camera or any live Beauty AR asset. The tutorial positions device classification as an application-specific input, encouraging teams to base tier assignments on measurements from their own supported hardware rather than generic benchmarks.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Five Questions to Ask Before Trusting an AI Agent's Fast Safety Judge

As AI agents increasingly automate tasks like sending emails and running database migrations, a common pattern has emerged where a small, fast model acts as a real-time gatekeeper, approving or rejecting each action in milliseconds. Drawing on research by Daniel Kahneman and Gary Klein, the article argues that such 'gut-call' models only deserve trust when they operate in predictable environments and receive clear, timely feedback on their decisions. Most agent setups lack this feedback loop, meaning the judge never learns whether its approvals led to good or bad outcomes. The piece outlines five critical questions teams should ask, covering data privacy, whether confidence scores are statistically calibrated, how uncertainty is handled, auditability of decisions, and vulnerability to prompt injection via adversarial inputs. The core warning is that a fast safety model with no error correction is not a reliable safeguard but a confident, unchecked gatekeeper.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

KVCoders fixes subtle grading bug that penalised correct Python output predictions

KVCoders, an online practice platform for CBSE Class 11–12 Computer Science students, discovered that its Output Practice mode was incorrectly marking students wrong due to minor spacing differences in predicted output. A plain string comparison flagged answers like [1,2,3] as incorrect even when the student understood the value, because Python's own print() adds spaces after commas. The fix involved building a character-by-character parser that normalises spacing only around Python container punctuation — commas and colons inside brackets — while leaving all string literal content completely unchanged. This ensures that outputs where whitespace is meaningful, such as formatted strings, are still graded strictly. The shared normalisation function is used across the web, Android API, and client-side preview endpoints to prevent inconsistent grading across platforms.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Graph-Based Fraud Tool Argues Both Sides Before Reaching a Verdict

A fraud investigation system called Fraud Investigator has been built on TigerGraph to address a core flaw in conventional fraud detection: risk scores alone both wrongly flag innocent users and miss real fraud. Data analysis of five source files revealed that all 900 cleared cases scored 0.81 or higher, while confirmed fraud cases had a mean score of 0.47, with 31% scoring below 0.3. The system runs separate prosecution and defence signature checks on each alert, then passes the evidence to a calibrated judge before a policy engine determines any real-world action. TigerGraph's graph features achieved an AUC of 0.91 in distinguishing alerts, compared to just 0.62 for the bank's own risk score. An LLM is used only for plain-language explanations and Suspicious Activity Report narratives, while all decisioning logic remains deterministic.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Windows error 0x80073D02 after auto-update? Here is how to fix it fast

Windows MSIX app updates fail with error code 0x80073D02 when background helper processes prevent the final package registration step from completing. Unlike the main app processes, these helpers — such as bundled CLIs, Node.js servers, or Python processes — carry no package identity and are invisible under Task Manager's Processes tab, requiring the Details tab to locate them. Some apps also install an auto-starting packaged service, like Claude Desktop's CoworkVMService, which can restart mid-repair and cause the registration to fail again. The fix involves identifying and killing all leftover processes by executable path rather than name, stopping any associated service, and then manually re-registering the package using Add-AppxPackage against the existing AppxManifest.xml. This approach typically resolves the issue in minutes without requiring a full reinstall.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

INKSHIFT lets event organizers move tables without losing guest bookings

A developer built INKSHIFT, a web app for managing games nights, workshops, and club meetups, as part of the Sanity Challenge. The tool lets organizers upload or type a plan, generate a shared signup link, and allow guests to book spots without creating an account. When a venue detail changes — such as a table becoming unavailable — organizers can review the proposed edit against existing bookings before approving it. Guests are automatically moved to the new location and retain their reservations without needing to re-register. The app uses Sanity's Content Lake, Workflows, and App SDK to keep sessions, spaces, and registrations linked throughout any changes.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

FinePrint Agent Cross-Checks Hackathon Rules Against Your Entry Details

FinePrint is an AI agent built for the Sanity Challenge that helps participants verify their hackathon entries against official contest rules. It reads a Sanity Knowledge Base constructed from structured requirement records, each tied to a source version, and checks user-provided facts against those rules with exact quotes. The tool highlights conflicts between rule documents — for instance, one page allowing unlimited entries while another caps submissions per path — and flags such cases as 'Rules unclear' rather than guessing. Users can also import any public rules page by URL, with the agent auto-proposing requirements that must be verbatim quotes or they are discarded. Currently covering 19 Sanity and 18 GIBC requirements, FinePrint saves reviews locally and generates downloadable Markdown reports, keeping imported event data in the browser rather than the shared dataset.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Claude Code's 'Compaction' Silently Erases Session Decisions — Here's the Fix

Claude Code automatically summarizes long sessions when the context window fills up, a process called compaction, which discards specific decisions and their reasoning while retaining only a general task summary. Users often don't notice the gap immediately because Claude continues responding coherently, but it may later redo work that was already decided against. A developer identified this as a key pain point after reader feedback on an earlier post about session handoffs. The proposed fix involves storing decisions outside the conversation so they can be reloaded automatically after each compaction event. ContextForge MCP version 0.12.0 implements this via a Claude Code hook that runs a recall command at session start, injecting the 10 most recent saved project decisions back into context right after compaction occurs.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer replaces placeholder portfolio with live data from four APIs in one week

A software developer rebuilt their portfolio site to display real-time data by integrating four external services within a single week. GitHub's GraphQL API was used to surface pinned repositories, while Hackatime provided live coding-hour statistics and language breakdowns. A headless CMS called Sanity replaced hardcoded blog posts, enabling content publishing without code changes or redeployments. Email delivery was activated through Resend, allowing contact form submissions to reach the developer's actual inbox via a custom domain. Key lessons from the project included the critical role of API response caching, the difficulty of silent CORS errors, and the long-term time savings from build automation.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Researcher Argues Memory Mechanisms, Not Scaling Laws, Are LLMs' Next Frontier

A researcher and developer has argued that capability gains in large language models stem from paradigm shifts in understanding intelligence — such as the Transformer and chain-of-thought reasoning — rather than scaling laws alone. The author contends that both breakthroughs essentially mimic higher human cognitive functions, with the Transformer simulating semantic understanding and CoT replicating explicit logical reasoning. Looking ahead, the author proposes that giving LLMs a human-like memory mechanism — distinct from the attention mechanism — is the critical next step. The argument draws on how human memory retains fuzzy state information rather than full detail, enabling reasoning and long-range control, a capability current attention windows cannot replicate. The author conducted a 3-million-token recall experiment to support the case that memory, not expanded context windows alone, is what LLMs fundamentally lack.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Your TLS Fingerprint Exposes Your HTTP Client Before Any Header Is Sent

A developer tested HTTPS requests to the 500 most popular websites from eight countries and found that one in five blocked the requests regardless of location. Investigation revealed that servers can identify a client's software long before reading HTTP headers, through a TLS fingerprinting technique called JA3 or JA4. The TLS ClientHello message — sent before any header — encodes details like cipher suite order, extension count, and supported protocols that differ distinctly between browsers and HTTP libraries. For example, Python's urllib advertises 30 cipher suites and only HTTP/1.1 support, while Chrome advertises 15 suites and HTTP/2, producing entirely different fingerprints. This means a script mimicking a browser via a spoofed user agent can still be detected, since the underlying TLS and HTTP/2 framing behavior gives away the true client.

0 comments Read more at DEV Community
← NewerPage 954 of 3187Older →