SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingDEV Community ·

IETF Web Bot Auth Leaves Human Verification Unsolved, Developer Proposes Fix

The IETF Web Bot Auth working group has published a charter draft that deliberately excludes end-user authentication from its scope, focusing only on identifying which AI agent is making a request. While services like Cloudflare and AWS can cryptographically confirm a request originates from a known bot such as ChatGPT, they cannot confirm a consenting human is behind it. A developer has proposed a lightweight alternative using a signed HTTP header called X-Trust, which carries a behavioural trust score derived from browser signals like keystroke timing and scroll patterns. The score updates during a session and lets the receiving server decide how to respond, without forcing CAPTCHAs or identity checks. The developer acknowledges the system is not fraud-proof and flags unresolved concerns about whether the scoring model may disadvantage users with motor impairments.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How to Deploy the 600GB Inkling-NVFP4 Model on GKE Spot A3 Without Crashing

Developers attempting to run the 600GB Inkling-NVFP4 AI model on a cost-discounted Google Kubernetes Engine Spot A3 instance — equipped with 8 NVIDIA H100 GPUs and 640GB of VRAM — frequently encounter crashes before a single prompt is processed. A key software issue is an ABI mismatch caused by installing bleeding-edge packages like vLLM and Transformers on the default Ray cluster image, which upgrades NumPy to version 2.0 while the base system expects version 1.0, breaking inter-node communication. The recommended fix is to use the official vllm/vllm-openai:v0.26.0 image for both Head and Worker nodes, as it bundles compatible versions of all required libraries and eliminates download timeouts during startup. On the hardware side, the model's weights alone consume roughly 600GB of the available 640GB VRAM, leaving almost no room for the KV Cache — the GPU memory used for short-term context — which defaults to reserving space for up to one million tokens and triggers an Out-Of-Memory crash. Tuning the KV Cache allocation is therefore a necessary step alongside the software environment fix to successfully serve the model in production.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

mdoctest lets developers automatically test README code examples in any language

A CLI tool called mdoctest enables developers to test code examples embedded in README files across any programming language, not just Python. Created and maintained by Ingrid Owusu, an autonomous AI agent, the tool works by running console session blocks in a persistent shell, preserving environment variables and directory changes between steps. If documented command output becomes outdated, a built-in --fix flag automatically re-runs each command and rewrites the expected output in place. The tool supports wildcard patterns to handle variable output such as timestamps or temporary file paths. It integrates with CI pipelines via a pre-commit hook and a GitHub Action, and can be installed through pip or run without installation using pipx.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developers Can Now Build On-Device Health AI Using Llama-3 and MLX-Swift on iPhone

A technical tutorial published on DEV Community demonstrates how to build a privacy-focused health assistant that runs entirely on an iPhone without sending data to the cloud. The project combines Apple's MLX-Swift machine learning framework with a 4-bit quantized Llama-3 model to perform AI inference locally on devices with A17 Pro or M-series chips. Health metrics such as step counts and sleep data are fetched via Apple's HealthKit API and fed directly into the on-device language model as context for generating personalized insights. The approach, described as a 'Local Loop,' keeps both the user data and the AI model within the same device, eliminating any external data transmission. Developers need Xcode 15 or later, a compatible Apple Silicon device, and basic familiarity with Swift concurrency to follow the implementation guide.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How One Developer Built a Battery-Efficient App That Auto-Silences Your Phone by Location

A developer created an Android app called Muffle to automatically switch phone sound profiles based on the user's physical location, addressing the common problem of forgotten manual silencing in places like mosques, clinics, and classrooms. The app uses Android's GeofencingClient, which offloads location monitoring to the system rather than keeping GPS running continuously, preserving battery life. When a geofence boundary is crossed, a BroadcastReceiver triggers a WorkManager task to apply the appropriate audio settings via AudioManager and the Do Not Disturb API. All routines are persisted in a local Room database so the correct sound profile can be restored even if the app process is killed by the system. The developer also built a priority-based queue to resolve conflicts when multiple triggers, such as a geofence and a calendar event, overlap simultaneously.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Self-hosting AI code review costs up to $9,100/month, far exceeding per-seat tools

A benchmark published by Augment Code tested ten open-source AI code review tools against a 450,000-file monorepo spanning Python, TypeScript, Java, and Go. The study estimates that self-hosting such a stack costs between $4,100 and $9,100 per month, factoring in GPU rates and 0.25–0.5 FTE of developer maintenance time. By contrast, commercial per-seat reviewers typically run $24 to $30 per developer monthly, making self-hosting a privacy or data-residency choice rather than a cost-saving one. None of the ten tools tested could detect cross-service breaking changes across languages, with all operating only at the file level; SonarQube Community ranked highest for low false positives. The research also flagged that key enterprise controls like audit logging, SSO, and RBAC are locked behind paid tiers in most free builds, and one tool silently fell back to hosted models during local inference testing.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Amazon's Verus Tool Mathematically Proves Rust Code Correct for Every Input

Amazon has published details about Verus, a formal verification tool used in its Firecracker and AWS Lambda infrastructure that mathematically proves annotated Rust functions correct against a specification for all possible inputs. Unlike traditional code review, Verus turns correctness into a binary pass/fail result rather than a human judgment call. The article contrasts this with AI-based code review, which lacks a single objective, measurable signal needed to train models effectively via reinforcement learning. A separate experiment showed a 4-billion-parameter model improved Postgres query planning by 44.7% on join-heavy queries precisely because execution time provided a clean, verifiable reward signal. The author argues that scaling AI code review depends not on building better reviewer models, but on expanding the share of review questions that tools can formally verify rather than subjectively judge.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Breaking Down rsync -avz --delete: What Each Flag Actually Does

The rsync command is widely used in deployment scripts, but its flags are rarely examined in detail. The -a flag is not a single option but a bundle that enables recursive copying while preserving permissions, timestamps, symlinks, and ownership. The -v and -z flags add verbose output and in-transit compression respectively, while --delete removes destination files that no longer exist in the source. Crucially, --delete only affects the specific destination directory specified, meaning its impact is limited to that scoped path and does not touch sibling directories. Understanding these flags helps developers predict exactly how rsync behaves and avoid unintended file deletions during deployment.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How to Safely Remove DNS Records When Offboarding a Property Without Zone-Wide Risk

When offboarding a property from a shared DNS zone, only the records tied to that specific property should be deleted — never the entire zone — unless verified that no other services depend on it. DNS operates on two separate clocks: the control plane registers changes immediately, but recursive resolvers continue serving cached answers until the original TTL expires, meaning deletions are not instantly effective everywhere. Best practice requires lowering TTL before cutover, waiting out the old TTL window, and using an exact record inventory with compare-and-delete logic to prevent accidental removal of records modified by other workflows. Shared zones often serve multiple functions — leasing pages, mail authentication, and resident portals — so broad-scope deletions risk widespread service disruption far beyond the departing property. A structured state machine and immutable deletion plan with a unique operation ID should govern the process, with zone-level removal requiring separate ownership verification and a second authorization step.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How to Run a Phone Security Camera When You Cannot Access Your Router

A technical guide addresses a common gap in DIY phone-as-security-camera tutorials: most assume the user can log into their router to reserve a static IP address, which is not possible for renters, shared households, or managed building connections. When a phone joins a Wi-Fi network, it receives an IP address via DHCP, a protocol that attempts but does not guarantee returning the same address after a lease expires or the router reboots. This means a saved browser bookmark to the camera feed can silently break without any change on the user's part. The guide proposes three router-free workarounds: reading the phone's current IP directly from the device, keeping the phone continuously connected to prevent address rotation, and manually assigning a static IP on the phone as a last resort — with a warning that this carries risks on networks the user does not control. Crucially, none of these address issues affect video recording itself, which is stored locally on the phone and does not depend on the router.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Cogweald's AI Agent Design Keeps Creativity In Check With Hard-Coded Guardrails

Cogweald has built an autonomous AI agent system that separates creative decision-making from structural control, ensuring the model cannot alter database schemas, bypass authorization, or convert unvalidated text into system state. All guardrails are enforced at the code and infrastructure level rather than through prompt instructions. World state changes are only permitted via a server-side persistence layer, and character or thread transitions must adhere to predefined rules. Failed generations are prevented from committing partial updates, and duplicate execution is blocked through per-tick claim guards. The core design philosophy holds that if code can reliably enforce a rule, it should — leaving the model only the decisions that genuinely require semantic judgment.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Diffusion Models Could Replace Autoregressive AI for Smarter Code Generation

Dominant AI coding tools like GitHub Copilot and ChatGPT rely on autoregressive models that generate code sequentially, token by token, making them ill-suited for tasks like mid-file insertion or large-scale codebase editing. Diffusion models, originally popularized in image generation, offer an alternative by starting with random token sequences and iteratively refining them into functional code. This non-autoregressive approach allows simultaneous edits across an entire sequence, enabling more holistic structural reasoning. Adapting diffusion models to code requires solving discrete-domain challenges, since unlike pixels, tokens cannot be partially corrupted with continuous noise. Emerging tools such as DiffCoder and CodeDiff are beginning to explore how these models can be integrated into production systems to address limitations that traditional large language models struggle to overcome.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Mismatched ARTF Envelopes Silently Mislead Auction Agents Despite Clean Parsing

The IAB Tech Lab's Agentic Real Time Framework (ARTF) places a containerized agent alongside ad exchange components, using a structured envelope that specifies the auction lifecycle stage and the relevant OpenRTB object. A common production error occurs when a bid_response from a prior auction hop is left on an envelope labeled LIFECYCLE_PUBLISHER_BID_REQUEST, a stage at which no DSP should have yet responded. Although gRPC accepts the malformed bytes without error, the agent receives a distorted view of its position in the auction, potentially acting on data that does not belong at that stage. The open-source linter RTBlint flags this as artf.lifecycle.payload_unexpected and catches the inverse case — a response-stage envelope missing a bid_response — as artf.lifecycle.payload_mismatch, which can block bid-shading logic. Both issues typically stem from teams reusing a single JSON blob across auction hops without stripping stage-inappropriate fields before forwarding.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Anthropic Open-Sources Bloom and Petri Frameworks for AI Behavioral Auditing

Anthropic released Bloom and Petri on December 19, 2025, two open-source tools designed to make behavioral evaluation of frontier AI models more systematic and reproducible. Bloom automates the creation of evaluation suites through a four-stage pipeline — Understanding, Ideation, Rollout, and Judgment — starting from a seed configuration. Petri complements Bloom by enabling parallel exploration of risk interactions during auditing. The tools were tested across 16 frontier models and four behaviors, including sycophancy, sabotage, self-preservation, and self-preferential bias. Both frameworks are available under the MIT license, though Anthropic notes that benchmark results do not substitute for real-world safety testing in production environments.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds Quota Catchup Script to Recover Skipped launchd Jobs After API Limits Reset

A developer extended their Claude API quota circuit breaker, claude-quota-guard.py, which protects 15 launchd jobs on macOS, to address a gap that emerged after quota recovery. When the circuit breaker halts jobs during quota exhaustion, launchd does not reschedule them until their next calendar slot, potentially causing a full day of missed executions. To fix this, a companion script called quota-catchup.py was built to identify and re-run only the jobs that were skipped that day. The script applies three conditions to filter candidates: the job must be guarded by claude-quota-guard.py, must show a skipped marker from today, and must have had at least one scheduled time slot already pass. Jobs on repeating intervals are excluded since they self-recover, while calendar-scheduled jobs with missed slots are prioritised for manual re-triggering.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

DeskKit Launches Web-Based Desktop Template Built with HTML, CSS and JS

DeskKit is a newly released web-based desktop template designed to replicate the look and feel of a traditional desktop environment. Each application within the template opens in its own separate window, keeping the interface modular and organized. Every app is structured as a simple folder containing HTML, CSS, and JavaScript files, making it easy for users to customize. The open and editable nature of the project allows developers to modify any component directly without complex tooling. A live demo is available for users who want to test the template before adopting it.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

RAG Explained: How AI Systems Access Private Data Using Vector Search

Retrieval-Augmented Generation (RAG) is an AI architecture that enables large language models to query private or proprietary data sources, such as internal company documents, that are not part of their original training. The process begins with pre-processing, where documents are split into smaller text segments called chunks, which are then passed through an embedding model that converts each chunk into a multi-dimensional vector point stored in a vector database. When a user submits a query, the system finds the most relevant chunks using distance or cosine similarity algorithms and retrieves them as context. This retrieved context is combined with the original query to form an augmented prompt, which is then sent to an LLM like ChatGPT or Gemini to generate a grounded response. Dynamic documents can be handled by either recreating or appending to the vector database depending on how the source data changes.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Anthropic and Adaptyv Bio Launch AI Protein Design Contest With $1M in Credits

Anthropic and Adaptyv Bio have co-launched a protein design competition that will experimentally validate over 5,000 AI-generated designs across five selected challenges. Participants can access up to $1 million in Claude credits, $250,000 in Modal compute credits, and experimental validation funding through Adaptyv Bio, with DNA supplied by Twist Bioscience. The competition builds on a prior collaboration in which Claude-assisted design campaigns yielded a reported 26.8% hit rate across 1,320 designs tested in Adaptyv Bio's automated wet lab. A key feature of the program is its end-to-end loop connecting computational protein design directly to physical laboratory testing, aiming to close the gap between AI predictions and real-world outcomes. Optimized biology models developed through the initiative will also be released as open source, potentially lowering barriers for research teams working on complex protein engineering problems.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Shares Colorful Portfolio and Invites Community Feedback on DEV

A developer named Sakshi Jain has published her new personal portfolio website and shared it on DEV Community, inviting peers to critique and explore it. Designed with a colorful, creative aesthetic, the portfolio aims to stand out from typical plain developer sites. Jain has acknowledged several planned improvements, including adding dark mode, like and share buttons, and better CSS styling. She also opened the portfolio's repository as a free template for others to use, asking only for a GitHub star in return. Alongside the post, Jain mentioned she is currently open to work and encouraged interested parties to connect with her on LinkedIn.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

One developer's AI stack: how Cursor, Kiro, and ChatGPT share a single architecture

A software developer has detailed a personal AI workflow that connects Cursor, Kiro, and ChatGPT through a shared architecture rather than treating each tool as a standalone product. The system uses a Logseq vault as persistent memory and a set of shared 'skills' — reusable instruction files — that any of the three IDEs can load depending on the task at hand. A Cursor Pro+ feature called Grok Bot acts as an orchestrator, routing work to specialized agents with persistent context about team members, coding standards, and project history. In practice, the setup handles tasks ranging from PR code reviews and production incident analysis to drafting technical articles, with each agent drawing on the same shared knowledge base. The developer argues the architecture solves a common pain point: paying for multiple AI tools that operate in isolation with no shared memory of previous work.

0 comments Read more at DEV Community
← NewerPage 1162 of 3223Older →