SShortSingh.

Programming

0
ProgrammingDEV Community ·

Developer Builds Agentless Home Server Orchestrator with 100 Pre-Built Stacks and Local AI

A developer has released NjordDeploy, an open-source, agentless deployment orchestrator designed for self-hosted home lab setups on devices like Raspberry Pi and Proxmox. The tool runs entirely from a control machine over SSH, leaving no background agents or daemons on the target server, preserving its full resources for user applications. It includes over 100 production-ready service templates, supports both Docker and rootless Podman, and generates configurations using a fully local AI engine powered by Ollama. NjordDeploy evolved from an earlier project called PiSelfhosting and was built to address common pain points such as port conflicts, bloated management platforms, and cloud-dependent DevOps tools. All deployments produce standard docker-compose files, ensuring no vendor lock-in even if the tool itself is discontinued.

0
ProgrammingDEV Community ·

Developer Documents Three Expo-Supabase GitHub OAuth Failures and Their Fixes

A developer building an Expo mobile app with Supabase authentication encountered three separate failures while implementing GitHub login. The first issue stemmed from registering a web URL instead of a custom app scheme as the OAuth redirect, leaving the browser session unable to reopen the app after GitHub approval. The second bug involved mishandling the PKCE flow by discarding the code verifier before the callback exchange could complete. A third failure was caused by a typo in the redirect URL, breaking the match required between Supabase and GitHub OAuth settings. The developer has shared the complete working implementation, totalling under 100 lines of code, to help others resolve similar issues faster.

0
ProgrammingDEV Community ·

Developer Builds Mobile PR Review App to Stop Delaying Contributor Feedback

A developer built a mobile pull request inbox app after repeatedly delaying code reviews for days, causing contributors to lose interest and PRs to accumulate merge conflicts. The app, built with Expo and Supabase, uses GitHub webhooks to push incoming PRs to a phone-based inbox where an AI model summarizes each diff in plain language. Users can swipe right to approve or left to request changes, reducing a typical review from days to under two minutes. The entire infrastructure runs on free tiers across Expo, Supabase, Vercel, and GitHub, costing nothing to operate. The developer is documenting the project as part of an ongoing build-in-public series.

0
ProgrammingDEV Community ·

How Missing Test Declarations Caused Ten Unrelated Tests to Fail Correctly

A software developer added three new checks to a build system without completing the required declarations for each, triggering ten test failures across unrelated parts of the suite. The build system enforced a structural invariant requiring every check to have a defined oracle and a list of fault conditions, and the missing fields broke that property system-wide. A separate block of work was outright refused because it referenced print names not present in the system's closed vocabulary, preventing silent data divergence. The developer noted that an incomplete check with no declared failure mode would otherwise have passed silently, inflating 'all checks passing' counts without actually being verifiable. The incident illustrated how well-designed systems surface rule violations at the point where the rule is defined, not where it was broken, making gaps impossible to ship undetected.

0
ProgrammingDEV Community ·

How to Convert Postman Collections into MCP Server Tools for AI Applications

Developers can transform existing Postman API collections into Model Context Protocol (MCP) tools, giving AI clients structured access to backend capabilities. Because Postman collections often predate formal OpenAPI documentation, they serve as a practical starting point, already containing request paths, parameters, headers, and authentication details. However, raw collections must be cleaned before conversion — removing secrets, duplicate requests, debug endpoints, and outdated API versions to avoid inheriting that messiness into the MCP server. Each Postman request is then mapped to an MCP tool definition, with path variables becoming required inputs and query parameters becoming optional ones, each carrying clear descriptions to help AI clients distinguish between similar identifiers. The process also requires decisions around authentication, error handling, schema validation, and production readiness before the tools are exposed to AI consumers.

0
ProgrammingHacker News ·

Jimmy Kimmel Interviews Texas Lawmaker James Talarico

Texas state lawmaker James Talarico appeared in an interview with late-night host Jimmy Kimmel, according to a link shared on Hacker News. The video was posted to YouTube, though no further details about the interview's content were provided in the source material. The submission received 12 points on Hacker News with no comments at the time of aggregation. Insufficient detail is available to summarize the substance of the conversation.

0
ProgrammingDEV Community ·

Developer builds unified semantic API for Brazil's fragmented official economic data

A developer created Open Economics, a semantic layer that consolidates data from multiple Brazilian official sources — including the Central Bank, IBGE, and Tesouro — behind a single API. The project's core challenge was not technical integration but preserving economic meaning, ensuring queries like 'current Selic rate' resolve to the precise correct series rather than a superficially similar one. The system maintains full data provenance in every response, recording the source institution, dataset identity, units, and any transformations applied. A remote MCP server with 19 tools was built to allow AI agents to query the data directly, but testing revealed a client-side interoperability issue where structured results were not being exposed to the underlying model. The fix involved adding a bounded plain-text representation alongside the structured output so agents could actually read and use the retrieved values.

0
ProgrammingDEV Community ·

macOS TCC Blocked App Despite Full Disk Access Due to Invalid Code Signature

A Flutter-based macOS disk utility called Helm was silently denied Full Disk Access by macOS despite the permission being explicitly granted in System Settings. Investigation revealed that macOS TCC validates an app's full code signature bundle integrity, not just its name or permission entry, before honoring privacy permissions. A codesign verification check exposed a mismatch between the sealed App.framework recorded at build time and the actual framework present in the bundle. The mismatch was traced to an Xcode build system crash that caused Flutter's embed step to run while the app's re-signing step was skipped, leaving a framework from a different build inside the signed bundle. The developer fixed the issue by re-signing the bundle and updated the release script to always verify code signatures on every build path, regardless of whether a Developer ID certificate is present.

0
ProgrammingHacker News ·

Open Letter Urges Anthropic CEO Dario Amodei to Release Claude Model Weights

A public open letter addressed to Anthropic CEO Dario Amodei has been published, calling on him to release the weights of the company's Claude AI models. The author argues that if Amodei genuinely believes in the principles he publicly espouses around AI safety and openness, he should back those words with action by open-sourcing the weights. The letter has gained traction on Hacker News, accumulating points and sparking discussion in the AI community. The debate reflects a broader ongoing tension between open-source AI advocates and companies that keep their model weights proprietary for safety or competitive reasons.

0
ProgrammingDEV Community ·

Screenshots Fall Short as Agent Review Artifacts, Structured Evidence Bundles Offer Fix

Screenshots shared in pull requests for AI agent runs lack critical context such as trace provenance, redaction details, and tamper evidence, making meaningful code review difficult. A more reliable alternative is a derived, redacted evidence package that includes structured files like HTML reports, JSON manifests, redaction reports, and hashed trace copies. The AgentInspect Evidence v2 bundle implements this pattern by separating source hashes from packaged-file hashes, ensuring reviewers can verify integrity without accessing private source traces. An offline verifier checks the manifest structure, file hashes, and provenance, flagging any post-finalization edits or unexpected files. This approach gives reviewers a bounded, inspectable artifact while keeping the original trace data private to the developer's machine.

0
ProgrammingDEV Community ·

AI-Generated Websites Accused of Looking Cheap and Identical, Like 90s Word Art

Frontend developers and designers are drawing comparisons between today's AI-generated websites and the low-effort Word Art aesthetic of the 1990s, arguing that users instinctively recognise AI designs as generic and cheap. A 2025 study found that over 80% of AI-generated websites share nearly identical underlying structures, as these tools largely replicate existing internet trends rather than producing original work. Tools like Framer's Wireframer can produce responsive layouts in under 60 seconds, but critics say the results look assembled from outdated SaaS design templates. Industry observers argue that AI primarily threatens the mid-tier template market, not skilled designers, since strategic thinking, branding, and SEO remain beyond its current capabilities. Agencies also warn that rebuilding poorly performing AI-generated sites often ends up costing more than hiring an experienced designer from the outset.

0
ProgrammingDEV Community ·

How a 1994 Compression Algorithm Became the Backbone of LLM Tokenization

Modern large language models rely on a tokenization method rooted in Byte Pair Encoding (BPE), a data-compression algorithm originally developed by programmer Philip Gage in 1994. Tokenization is the process of breaking text into smaller units called tokens before a model processes them, and the way this split occurs affects input length, inference cost, vocabulary size, and multilingual performance. BPE works by repeatedly merging the most frequently co-occurring byte sequences in a training corpus, striking a balance between character-level and word-level approaches. Unlike word tokenizers, BPE can handle rare or invented words by decomposing them into known sub-units, without requiring an infinitely large vocabulary. The algorithm gained widespread adoption in neural machine translation before becoming central to tokenizers used in models like GPT-2 and its successors.

0
ProgrammingDEV Community ·

HexaLayered Architecture Gets AI-Enforced Rules via Claude Code Integration

A developer has extended the HexaLayered Architecture project by integrating it with Claude Code, an AI coding assistant, to automatically enforce architectural rules rather than relying on manual code reviews. A new configuration directory containing a 'constitution.md' file defines strict, testable rules about class visibility and layer boundaries within Spring Boot projects. A shell script runs on every file write and exits with an error code if any architectural violation is detected, sending feedback directly to the AI model in real time. The toolkit also introduces a Spec Driven Development flow with commands that guide planning and class placement before any code is written. The setup can be copied to a new project with a single command, and a bootstrap process automatically detects the base package, build tool, and test configuration.

0
ProgrammingDEV Community ·

A Developer's Guide to Choosing the Right PDF Splitting Strategy

Splitting a PDF into multiple output files becomes complex at scale, requiring engineers to think carefully about how page ranges map to output containers. The PDF format stores pages as a tree of indirect objects, meaning logical page order can differ from raw file index, which naive splitting tools often mishandle. Engineers can rely on three core strategies: fixed-size partitioning for predictable batches, boundary-driven partitioning for documents with structural markers, and selector-driven partitioning using a manifest file for audit or legal work. A practical rule of thumb is to use fixed or boundary-based splits when input data changes but rules stay constant, and manifest-driven splits when the mapping itself varies per batch. Documenting the page-to-file mapping explicitly before processing is recommended to avoid hard-to-trace errors in output validation.

0
ProgrammingDEV Community ·

Developer fixes false-stale alerts by replacing byte pins with key-based anchors

A software team maintained a code citation register where rows turned 'stale' whenever their referenced bytes changed, but legitimate actions like adding a new layer were triggering nine false alarms at once. The tempting fix — repinning to refresh the hash — was ruled out because it would launder real staleness into green alongside false positives, making the tool unreliable. Investigating the root cause revealed that the affected rows only needed to confirm a declaration existed and was findable, not track its exact byte contents, which three separate parsers already validated. The team introduced a new 'key anchor' type that checks for the presence of a named key rather than byte-level content, eliminating positional drift and making the repin command irrelevant for those rows. Ten rows migrated to the new anchor type while 124 retained byte pins, and the test suite expanded from 74 to 101 arms to ensure both anchor types behave correctly under real changes.

0
ProgrammingDEV Community ·

TryHackMe Metasploit Payload Generation Room: Key Answers and Walkthrough

A writeup for TryHackMe's Metasploit Payload Generation room covers core concepts including stageless versus staged payloads, where stageless payloads are self-contained and require no secondary download. The guide walks through msfvenom commands for generating Windows and Linux payloads, covering flags for output format, bad character exclusion, and executable template injection. It also explains configuring a universal listener using the exploit/multi/handler module, with ExitOnSession disabled to maintain multiple sessions. The capstone task involves uploading a reverse shell via an SMB share, then using Meterpreter to dump NTLM hashes and retrieve a flag from the Administrator's Documents folder.

0
ProgrammingHacker News ·

Linux Zoom Client Found Silently Reading All X11 Clipboard Contents

The Linux version of the Zoom desktop client has been observed proactively reading everything written to the X11 clipboard, even when the application is not in active use. This behavior was flagged by Simon Tatham, a well-known developer, who shared his findings on the Mastodon platform. The clipboard on X11 works differently from other systems, potentially making such passive monitoring easier to implement. This raises significant privacy concerns, as sensitive data copied by users could be accessed by the Zoom process without their knowledge or consent. The issue was discussed on Hacker News, drawing attention from the developer community.

0
ProgrammingDEV Community ·

Developer finds 374 of 412 privacy scan hits traced to one repeated line of build tooling

A developer running a pre-publication privacy scan on a code repository flagged 412 hits across 16 files, initially suggesting widespread issues requiring extensive review. Closer analysis revealed 374 of those hits were the same single line — a build tool emitting an absolute home directory path — repeated across 208 logged control runs. Rather than resetting the append-only log to hide the leak, the developer chose to exclude the private log from publication and start a fresh public log using a relative path token, preserving the integrity of both records. The generator was fixed to emit a repo-relative token instead of an absolute path, resolving the root cause with one code change. The incident highlighted that build tools and test harnesses, unlike application code, are rarely reviewed for what they write into permanent logs.

0
ProgrammingDEV Community ·

Laid-Off Automotive Software Architect Reflects on Career Shift and Industry Decline

A software architect with nearly a decade of experience in the automotive embedded systems industry was laid off by end of 2025 as part of company-wide redundancies. Rather than aggressively job-hunting, he used the time to revisit foundational hardware platforms and explore new technical skills at his own pace. He found that his background in C-heavy embedded development made transitioning to desktop or systems C++ roles difficult due to a lack of formal commercial experience in that area. Observing peers accept steep pay cuts or search fruitlessly for openings, he concluded that the traditional automotive software market is unlikely to recover and that the broader software engineering job market is shrinking. He now plans to pursue a hybrid livelihood combining handyman work, freelancing, and small local technical projects, having stepped away from the conventional IT career path.

← NewerPage 1410 of 3335Older →