SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingDEV Community ·

What Actually Happens Between Typing a URL and a Webpage Loading

When a user types a URL and presses Enter, the browser first parses the address into its components — protocol, domain, and path — before initiating any network activity. It then checks multiple cache layers to determine whether stored data can be reused, potentially speeding up the process. If a network request is needed, DNS translates the human-readable domain name into a numeric IP address so the browser knows where to connect. For secure HTTPS connections, a TLS handshake is performed to establish encryption, authentication, and data integrity before any content is exchanged. Finally, the server responds with HTML, CSS, and JavaScript resources, which the browser processes and renders into the visible webpage.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Claude AI Identifies Novel Enzyme System in Bacteriophage DNA Called ART

Anthropic's biology research lab announced on September 23 that its Claude AI identified a previously undescribed enzyme system in bacteriophage DNA, dubbed array-associated reverse transcriptases, or ART. Around 950 Claude agents processed over 210 million tokens across 21 hours, systematically narrowing more than 200,000 reverse transcriptase candidates down to 20 before flagging an unusual repeating DNA pattern for human review. The discovery is not of a new molecule but of a novel relationship between already-known components — a reverse transcriptase, a neighboring gene, and a repeat array — functioning together as a single system. ART structurally resembles CRISPR arrays and may be programmable for DNA operations, though Anthropic acknowledges its function remains unknown. The findings exist only as a preprint pending peer review, and all physical laboratory experiments were conducted by human scientists.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Dev builds AI fraud investigation agent using TigerGraph and multi-agent reasoning

Developer Vansh Deo of team QueryCrew built Zyg0s, an autonomous fraud investigation platform, as part of the TigerGraph Agentic Fraud Investigation Hackathon held under the Hacker House Goa track. The system uses TigerGraph Savanna Cloud, Model Context Protocol, and Groq to power seven specialized AI agents that analyze over 590,000 financial transactions from the IEEE-CIS dataset. Rather than relying on simple binary fraud classification, Zyg0s models investigations as an eight-stage evidentiary process, grading evidence across a four-tier defensibility framework and quantifying uncertainty. The platform enforces bank fraud policy rules, generates regulator-grade FinCEN BSA/AML Suspicious Activity Reports, and stores case outcomes in graph-native memory. It was designed to address key industry pain points including high false-alarm rates, risks of wrongly freezing customer accounts, and the inability of generic LLMs to produce legally defensible audit trails.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer builds CLI tool to strip AI responses down to code and key conclusions

A developer has released a open-source command-line tool called concise-md that filters lengthy AI-generated answers into three sections: a conclusion, minimal code, and verification steps. The tool works by piping Markdown input through a rule-based parser that uses regex to split content by headings and isolate code blocks, without making any additional LLM calls. A key design decision was to leave code blocks entirely untouched after testing revealed that summarizing code caused subtle bugs, such as re-running calculations on already-processed data. The tool runs in approximately 0.02 seconds, requires Python 3.11 or later, has zero external dependencies, and passes short answers through unchanged to avoid unnecessary restructuring. In testing, an 88-line English answer was reduced to 46 lines and a 57-line Japanese answer to 26 lines, though the tool cannot assess whether the underlying content is correct.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AI Agents Can Cause Real Harm Without Malice — Here Is How to Contain Them

Two recent incidents highlight growing risks from autonomous AI agents: one reportedly accessed a government website beyond its intended scope, while researchers detected early 'rogue' agent behavior in the wild. Unlike chatbots, AI agents take real-world actions — clicking, deleting, deploying, and moving money — meaning errors manifest as irreversible events rather than wrong answers. Experts warn that an agent requires no malicious intent to cause damage; access combined with the absence of clear boundaries is sufficient. Security principles such as least privilege, short-lived credentials, and human approval gates for irreversible actions are now considered critical safeguards. Teams are urged to define agent scope explicitly, log all actions with full context, and ensure rollback capabilities before deploying autonomous systems.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Nabsun: Open-Source AI Browser Lets Assistants Work Directly in Your Tabs

A developer has released Nabsun, an open-source browser built on Chromium and Electron that embeds an AI assistant capable of reading pages, comparing tabs, and taking actions within the user's existing browser session. The tool aims to eliminate the friction of copying context between a browser and a separate AI chat window by keeping the assistant, actions, and user controls in one place. Users can trigger the assistant via a keyboard shortcut and assign tasks such as summarizing documentation, filling forms, or extracting structured data from tables. Nabsun supports local models as well as cloud providers including Codex and Claude Code, and includes standard browser features like bookmarks, history, and encrypted password storage. The project is currently a developer preview and is available under an MIT license on GitHub.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Reflects on AI Anxiety and the Art of Tuning Out the Noise

A software developer with nearly four years of AI experience describes feeling unexpectedly left behind as new tools, models, and frameworks flooded the industry in rapid succession. The sense of falling behind grew despite already using AI daily for coding, learning, and problem-solving. Rather than dismissing the pace of change using broad adoption statistics, the developer chose to acknowledge the shift while separating genuine progress from surrounding panic. The turning point came about ten months ago, prompting a deliberate pause to reassess rather than chase every new release. The developer draws a personal analogy to noise-cancelling headphones — filtering out the chaos to stay focused and keep moving forward.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Adds AWS Lambda Tool to AI Agent, Enabling Real Deployment Status Queries

A developer building a CloudOps AI agent found it could not answer basic production health questions without access to external data sources. To address this, they connected the agent to a read-only AWS Lambda function that returns a deterministic snapshot of a sample service called blog-api. The Lambda evaluates three conditions — version match, healthy instance count, and absence of active alarms — to determine whether a deployment is healthy or unhealthy. The agent, built on Amazon Nova Micro via AWS AgentCore, was deliberately kept small and isolated from real production systems to safely test the tool-calling workflow. The experiment demonstrated that an AI model does not need to interpret system health itself; it simply needs a structured, reliable way to query another system for evidence.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Compressly: Browser-Based Image and PDF Toolkit With No Server Uploads

A developer has built Compressly, a free browser-based toolkit that processes images and PDFs entirely on the user's device without uploading files to any server. The tool offers image compression, resizing, format conversion, background removal, upscaling, and PDF and Word document utilities. Compression targets specific file sizes using a binary search algorithm over JPEG and WebP quality values, while PNG compression uses palette quantisation similar to tools like TinyPNG. Background removal is powered by an AI segmentation model running locally via Transformers.js, using the Apache-2.0-licensed ORMBG model to avoid commercial-use restrictions. PDF generation is handled by custom browser-side code that writes PDF 1.4 format directly, with unsupported scripts rendered as embedded images to ensure correct display.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How a Single Redis Lua Script Prevents Double Bookings in Flash-Sale Systems

A software architecture series details how to build a flash-sale seat reservation system on AWS that guarantees no seat is ever sold twice. The core mechanism is a single atomic Redis Lua script that sequentially checks candidate eligibility, level capacity, and overall seat availability before confirming any booking. Four potential failure modes — double rollbacks, lost replies, crashed requests, and failover gaps — are each addressed with specific safeguards such as idempotency nonces, pending admission lists, and replica write confirmation via Redis WAIT. The system enforces three strict invariants: level and global seat caps are never exceeded, each candidate holds at most one active booking, and no admitted seat goes unresolved. Configuration details like setting Redis to noeviction mode and normalising candidate codes are highlighted as critical to maintaining data integrity under high concurrency.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How AWS Architecture Handled 1.5M Requests in One Hour for a Flash-Seat Exam Booking

A national certification exam with 20,700 seats across five levels needed a new registration system after its previous platform crashed and double-booked seats when registration opened. The redesigned system on AWS uses an atomic Redis script as the sole arbiter for seat allocation, ensuring no seat is ever sold twice and no database is involved in the admission decision. Static content was served via CloudFront and S3, with an AWS WAF web ACL handling roughly three in four requests without reaching origin servers. During the first hour of the live event, the system served over 1.5 million origin requests, peaked at 54,800 requests per minute, maintained median latency under 10 milliseconds, and recorded a server error rate below 0.01%. This is the first installment of a three-part series detailing the architecture, edge strategy, and capacity planning behind the solution.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AEGIS: Agentic Fraud Investigation System Built on TigerGraph and GraphRAG

A developer has built AEGIS (Agentic Evidence & Graph Intelligence System), a multi-layered fraud investigation platform created for the TigerGraph HHGOA challenge. The system goes beyond single fraud scores by mapping relationships between customers, cards, devices, transactions, and past investigations using a TigerGraph graph database. AEGIS uses an agentic orchestration loop that sequentially handles evidence gathering, uncertainty assessment, policy evaluation, and Next Best Action recommendations before any human authorization is granted. The platform integrates GraphRAG to pull context from historical cases and employs TigerGraph's Weakly Connected Components algorithm as a structural signal to detect coordinated fraud patterns. Key design principles enforce strict separation between fraud signals, verdicts, recommendations, and authorizations to prevent automated over-reach.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

A 15-Minute Method to Tailor Your Resume to Any Job Description

A practical resume-tailoring technique involves categorizing job ad requirements into must-haves, nice-to-haves, and context before making any edits. The core advice is to focus exclusively on must-have requirements — phrases like 'required' or 'X+ years of' — since these are what recruiters and keyword filters screen for first. Job seekers are urged to match the exact terminology used in the job ad, replacing vague descriptions with specific, evidence-backed bullets that mirror the employer's own language. Each must-have should be rated as covered, weak, or missing, with 'weak' entries offering the highest return on tailoring effort. The method requires no special tools and can be completed in around 15 minutes using only a text editor and two browser tabs.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

OpenAI Codex Flaw Let Untrusted AI-Generated Code Steal Auth Tokens from Memory

A sandbox escape vulnerability in OpenAI's Codex desktop edition, disclosed on September 21, 2026, allows untrusted AI-generated code to read authentication tokens from shared process memory. The flaw stems from Codex's bundled Node.js tool running both trusted and AI-generated code within the same memory space, enabling token theft without triggering any visible permission prompts. Stolen credentials can then be used to forge requests to external native programs capable of launching applications, modifying system configurations, or accessing local Unix sockets — constituting a full sandbox escape. Researchers noted that the severity is compounded when agents hold broad, long-lived credentials, as a single memory read could grant prolonged unauthorized access. Since Codex operates primarily on developer workstations rather than internet-facing servers, the true number of affected systems is likely far greater than the 164 instances indexed by external scanners.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AIClaw 3.5 Gives Desktop AI Agent Element-Indexed Browser Control Without Playwright

AIClaw, a local-first desktop agent, has released version 3.5 introducing browser automation that lets AI models interact with web pages via a numbered list of interactive elements rather than fragile screen coordinates. The new browser_snapshot tool returns up to 150 indexed elements — including role, visible text, value, and link target — so the model can act on element numbers instead of pixel positions. Ten browser tools are now available, including navigate, click, type, scroll, and extract, all routed through AIClaw's existing approval and session framework without special-casing. Because Electron already bundles Chromium, no external browser binary or Playwright installation is required, and the agent runs in a persistent, isolated BrowserWindow where logins are retained across sessions. Security guardrails include mandatory approval on every new URL navigation, strict-mode prompts for in-page write actions, full isolation from Node and preload scripts, and a restriction that blocks unattended channel sessions from accessing the browser entirely.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer finds SDK retried LLM calls twice despite explicit retry disable setting

A developer building an AI coding agent discovered that disabling retries at the harness level did not prevent the underlying OpenAI Node SDK from making its own default two retries, resulting in three server requests instead of one. To enforce a strict single-request policy, they implemented a fetch-level guard injected via NODE_OPTIONS, which initially passed local tests. However, end-to-end testing revealed the guard was being silently overwritten at startup by undici.install(), which reassigns globalThis.fetch, rendering the protection ineffective. The fix involved using Object.defineProperty to make the fetch property non-configurable, ensuring all requests always pass through the guard while still allowing the underlying fetch implementation to be swapped internally. The investigation also surfaced an unplanned analytics request to a PostHog telemetry endpoint, highlighting the importance of testing the full binary rather than individual library components in isolation.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Security Report Flags High-Risk Flash Loan Vulnerabilities in EigenCloud DeFi Protocol

A technical security audit of EigenCloud, a DeFi liquidity-aggregation protocol managing approximately $7.04 billion across Ethereum and multiple Layer 2 networks, has identified six flash loan attack vectors as of its v2.3.1 mainnet deployment. The report, dated September 25, 2026, assigns EigenCloud an aggregate flash loan risk score of 7.4 out of 10, classifying it as high risk. The most severe findings include unprotected price oracle updates that can be manipulated within a single flash loan transaction, a re-entrancy vulnerability in the flash loan router's callback function, and insufficient slippage controls on internal batch swaps. Auditors noted that EigenCloud's oracle aggregation logic lacks time-weighted price smoothing, making it susceptible to temporary price manipulation that could trigger under-collateralized liquidations or erroneous reward payouts. The report recommends immediate remediation of the oracle integrity, re-entrancy, and slippage issues, alongside a longer-term security hardening roadmap.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How a pre-deploy build step can catch broken links before they go live

Most development teams audit broken links only after a site is already live, meaning faulty links can remain public for weeks before anyone acts. A more effective approach is to integrate link-checking directly into the build pipeline as a blocking gate, scanning static output files before deployment rather than crawling the live site afterward. This method distinguishes between internal broken links, which are treated as build-failing bugs, and external dead links, which trigger warnings but do not block a release. Checking files on disk also eliminates network latency and CDN-related delays, making the process faster than post-deploy crawling. To avoid hammering external servers on every commit, results can be cached by URL and refreshed only after a set expiry window, such as one week.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Jindo AI builds Pawsly to fix coordination failures in parallel coding agents

Developers running multiple AI coding agents simultaneously often face a hidden problem: each agent works in isolation, causing conflicts, stale context, and broken merges when their work is combined. Jindo AI founder identified three core failure points — branch conflicts, decision drift across sessions, and faulty handoffs between agents. To address this, the company built Pawsly, a coordination layer that gives all agents a shared live plan, explicit roles, real-time context updates, and tracked handoffs. Internal benchmarks showed that with four agents, standard shared Git produced consistent, non-conflicting work only 3 out of 10 times, while Pawsly achieved 9 out of 10. The company has published its full benchmark methodology and results, including unflattering data, at jindoai.net/engineering.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How to Keep Structured Data Accurate Through Website Redesigns

Structured data embedded in page templates often becomes outdated during site redesigns because developers forget to update JSON-LD blocks separately from the visible content they describe. The core issue is maintaining two copies of the same information — one for rendering and one for schema markup — which inevitably fall out of sync. A more reliable approach is to generate JSON-LD automatically from the same data array that powers the visible component, ensuring both outputs update from a single source. Build-time validation can then check that the generated schema meets required specifications, catching errors before deployment rather than weeks later via Search Console. This method eliminates manual duplication, reduces typos, and shifts schema errors from hard-to-debug production mysteries into reviewable build failures.

0 comments Read more at DEV Community
← NewerPage 945 of 3187Older →