SShortSingh.

Programming

0
ProgrammingDEV Community ·

Why an AI write approval needs two timeout windows, not one

The article argues that a single timeout cannot answer two distinct questions when an AI requests permission to write and a person must approve it: whether the conversation should keep waiting, and whether the request can still be decided at all. The author describes a bug where two places computed the offline expiry boundary differently, so the pending list showed rows as actionable that had already expired, causing clicks to fail. The proposed fix separates a short in-conversation wait (60 seconds by default) from a long offline decision window (24 hours by default), and routes every boundary calculation through one shared function. It also addresses races between the in-conversation button and the pending list by using an atomic conditional database update that only succeeds if the status is still pending.

0
ProgrammingDEV Community ·

Capturing the request path to debug a 404 from the OpenAI Node SDK baseURL

A 404 from an SDK call does not reveal which URL the client actually requested, so the author suggests first checking the exact request path. The tutorial runs the OpenAI Node SDK against a local HTTP fixture on the developer's machine, requiring no model account or real API key. The fixture accepts only /v1/chat/completions and records the method, path, and model field for each request. Five baseURL configurations are tested, including the common mistake of including the full endpoint in baseURL. The author notes that a fixture 404 only shows a route mismatch in the test setup, not a provider outage or unsupported model, and tested with Node.js v26.0.0 and openai 7.38.0.

0
ProgrammingDEV Community ·

Generate Invoice PDFs From JSON in Node.js Without Templates or Puppeteer

The article shows how to create an invoice PDF by sending a JSON description to the GetInvoicePDF API, which returns a finished file. A short Node.js script with no dependencies posts the invoice details, including line items, tax rate and accent color, and saves the result as invoice.pdf. The API can alternatively store the PDF and return a private URL that requires the same API key to access. Customers can be sent the PDF, and once payment is received the same data can be resubmitted with a paid flag to produce a receipt stamped PAID. The author notes the approach is unsuitable for pixel-perfect custom layouts or legally structured e-invoice formats such as Factur-X, ZUGFeRD, PEPPOL and FatturaPA.

0
ProgrammingDEV Community ·

Hum to Song measures generated backing tracks instead of trusting the prompt

The Hum to Song project aims to ensure a hummed melody appears unchanged in the final song. Its pipeline transcribes the hum into notes, tempo, key and chords, then asks Google's Lyria text-to-music model for a matching backing band. Testing showed the model often ignored the requested tempo, key, chords and intro length. The team therefore stopped relying on the prompt and instead analyzes the returned audio in the browser. The analysis estimates tempo and downbeat positions, detects where the intro ends, and measures the backing's chords to determine any needed transposition, so the melody can be aligned with the band.

0
ProgrammingDEV Community ·

Uptime monitoring for experiments should separate reachability, job deadlines, and cohort outcomes

A green uptime aggregate can conceal damage to a single experiment cohort, so the article argues that rollback decisions need more than a global availability figure. It recommends three distinct signals: external probes for reachability, heartbeat deadlines for scheduled jobs, and cohort-tagged application outcomes that drive rollback. The author notes that a health endpoint cannot verify whether a new rule works for a specific cohort or whether jobs finished on time. To control metric cardinality and privacy exposure, the piece advises emitting only stable cohort names rather than raw tenant identifiers, and keeping identifiers out of URLs and public status pages.

0
ProgrammingDEV Community ·

Five open science challenges let hidden verifiers, not submitters, score AI-proposed answers

FINAL-Bench is running five open science challenges that separate proposing an answer from judging it. Participants submit items such as drug candidates, lithium electrolyte crystals, superconductor pairing estimates, quantum energies or decoders, and a verifier they cannot see scores each one. Answer keys and scoring implementations stay private on the organiser's hardware, while rules and weights are published, so entries cannot be tuned to a public grader. Challenges are open on staggered deadlines, and a LEADBOARD of 21 boards provides starter prompts and skeletons so no special hardware is needed. AI coding agents are explicitly allowed to participate. All scores are computational predictions rather than laboratory measurements, so high scores are hypotheses to be tested.

0
ProgrammingDEV Community ·

Risk report flags oracle manipulation vectors in Polygon Bridge, rated 7/10

A security audit dated 19 October 2026 assessed the oracle integration layer of the Polygon Bridge, which links Ethereum and Polygon and holds roughly $2.6 billion in assets. The researchers rated the overall risk as high, 7 out of 10, while noting the bridge's core functionality remains sound. The report identifies four high-impact attack vectors that could be exploited through flash loans, control of validator nodes, or influence over off-chain price data. Among them, stale oracle updates and compromise of a Chainlink aggregator could enable under-collateralized withdrawals or artificial liquidations. The authors estimate a worst-case coordinated exploit could cause up to $300 million in losses.

0
ProgrammingDEV Community ·

GitHub Adds Coverage Skip for New Branches and Launches Dynamic Copilot Workflows

GitHub changed its upload-code-coverage workflow so that pushes to new branches without an existing pull request now skip the upload instead of failing CI. Default branch and pull request events still upload coverage, and teams can add a pull_request trigger to get coverage sooner. Separately, dynamic workflows in Copilot CLI and the Copilot app entered public preview for all Copilot plans, letting developers write the steps, parallelism, and agent turns in code. The newsletter also covers the HydraFusion shortcut for routing work across patterns, along with a critique of 'model welfare' arguments about AI moral status.

0
ProgrammingDEV Community ·

Roll-Call Uses AI to Generate Daily Standups From GitHub and Linear Activity

Roll-Call is an open-source AI tool that automatically drafts daily standup updates by reading a team's GitHub pull requests and commits and its Linear issues. Each night it separates completed work from ongoing tasks and links issues to the PRs that implement them, then uses an open-weight Gemma model to write a plain-English update per team member. The reports are stored in MongoDB Atlas and uploaded to Backboard, where a chat agent can answer questions about the team's history. The project is aimed at small remote teams of roughly 5 to 20 people and at agencies managing client repositories. Its developer shared an unedited sample report that tracks the tool's own development.

0
ProgrammingDEV Community ·

Five Trending Developer Tools This October, from AI Git Automation to Native PC Ports

A DEV Community roundup highlights five developer tools gaining traction on GitHub in October 2026. Git-Designed lets AI agents manage code repositories autonomously and has drawn about 2,300 stars since its October 15 release. Qwen-Image 2.1 NSFW/Pose Skill adds content detection and pose estimation to AI-generated images, with roughly 2,800 stars. Scout is an open-source browser agent for web automation and data extraction, alongside BRUTAL, a free C# game framework offering direct access to native APIs and Vulkan. The list also features a native Windows PC port of Uncharted: The Nathan Drake Collection by SuperMedo. The author compiled the rankings from GitHub activity, Hacker News, web searches, and custom crawlers.

0
ProgrammingHacker News ·

Sanity's Edge: A Voice-Based BBS Experience Modeled on 1994

Sanity's Edge is a project presented on Hacker News that recreates the feel of a 1994-era bulletin board system, but through voice interaction. The item was shared via a link to the project's website and received 6 points. No comments had been posted at the time the item was captured. The source material provides little detail beyond the project name, its retro BBS framing, and its voice-based format.

0
ProgrammingDEV Community ·

Banks and Publishers Restrict AI Agents as Identity Gaps Emerge

After customers began handing banking credentials to AI assistants, Citigroup and Bank of America moved to bar agents from client accounts, and Schwab began blocking outside AI tools that attempt to move money. Content publishers showed a similar shift, with Cloudflare's one-click AI crawler block adopted by over a million customers and the share of reputable sites blocking AI scrapers rising sharply from 2023 to 2025. Analysis of crawler traffic suggests much of it is training-related rather than live user activity, while robots.txt enforcement remains inconsistent across top domains. The article argues that current agents cannot prove who they are or who authorized them, leaving operators to rely on easily spoofed signals. Proposed remedies include cryptographically signed agent identities under RFC 9421 and Visa's Trusted Agent Protocol for commerce.

0
ProgrammingDEV Community ·

AIClaw Adds Opt-In Skill Access Tracking That Excludes Prompt and Tool Content

AIClaw's v3.7.3 release introduced an opt-in OpenTelemetry Logs event, aiclaw.skill_invocation, that records when an agent accesses a skill and how that access was triggered. The event classifies access as explicit, when a user mentions a skill and it loads, or implicit, when a skill loads or its SKILL.md is read without such a mention. Prompt text, skill bodies, and tool payloads are kept out of the exported events, and file paths are resolved locally. The tracker deduplicates repeated loads within a turn, so one turn can produce both an explicit and an implicit event for the same skill. The feature was released October 8, and the latest public version at the time of writing is v3.7.15.

0
ProgrammingDEV Community ·

ElevenLabs AI Voice Helps Real Estate Sites Add Narrated Virtual Property Tours

A developer guide explains how AI text-to-speech can add narration to property listings, making virtual tours more immersive and engaging. It describes how to get started with ElevenLabs, a TTS service with a free tier, by generating an API key and listing available voices. The article includes a Python script that converts a property description into an MP3 narration file via the API. It also covers voice cloning as a way to keep a consistent brand voice across listings. The approach is presented as a low-cost alternative to studio recording and hired narrators.

0
ProgrammingDEV Community ·

Fixing Stripe webhook 401 errors on Supabase Edge Functions by disabling JWT verification

A common cause of Stripe webhooks failing with 401 on a Supabase Edge Function is the platform's built-in JWT check, which rejects requests before the function's code runs. The article explains how to tell this apart from a 401 returned by the developer's own code, by inspecting the response body in Stripe's delivery log. The fix is to disable JWT verification for the webhook function only, either via config.toml or the --no-verify-jwt deploy flag, so that redeploys don't re-enable it. Because the endpoint becomes publicly reachable once this change is made, the author stresses that the Stripe signature must then be verified inside the function, since otherwise anyone who knows the URL could send forged events.

0
ProgrammingDEV Community ·

Parallel Claude Code agents: why Git worktrees miss semantic conflicts

Running several Claude Code agents at once is far more manageable when each one gets its own Git worktree on a separate branch, which prevents them from overwriting each other's files. However, worktrees do not stop agents from producing plans that contradict each other, since Git merges text rather than intent and can combine such changes cleanly. These so-called semantic conflicts, such as a renamed function still being called elsewhere or two agents building the same feature twice, go unreported by Git. The article notes that worktrees also leave gitignored dependencies, ports, databases and branch checkouts unisolated, so each agent needs its own install and a distinct branch. It recommends assigning one integrator to merge branches and running real checks on the combined result rather than on each branch separately.

0
ProgrammingDEV Community ·

Developer Shares Gmail Agent Side Project Built for Hacktoberfest AI Challenge

A developer has published a Gmail agent built as a side project for the Hacktoberfest Open-Source AI Challenge Week 2. The submission, shared on DEV Community, includes demo code hosted on GitHub. The post walks through how the agent was built and touches on why open innovation matters. It also references the agent session and prize categories tied to the challenge, along with the Mastra and Backboard tools. The entry is tagged with the challenge's hashtags.

0
ProgrammingDEV Community ·

Fixing undefined or invalid API key errors from environment variables on Vercel

Apps that run locally can fail on Vercel with errors like missing Supabase URLs or invalid OpenAI keys, because git-ignored local env files are never deployed. The usual causes are variables added after the last deployment, variables set for the wrong environment, or missing public prefixes such as NEXT_PUBLIC_ or VITE_ in browser code. Variables can be added through the Vercel dashboard or CLI, but a redeploy is required because public prefixes are baked in at build time. Secret keys should never carry a public prefix and should be used only in server-side code, since anything prefixed for public use is exposed to every visitor. The article also advises checking for stray quotes or spaces in values and noting that a public Supabase anon key is only safe when Row Level Security is enabled.

0
ProgrammingDEV Community ·

Fixing Supabase Row-Level Security Errors in Lovable and Bolt Apps

Apps built with Lovable or Bolt on Supabase often hit Postgres error 42501 when inserting rows, or see empty results when reading, because Row Level Security is blocking the action. These failures typically occur when RLS is enabled without a matching policy, the user is not logged in so auth.uid() is null, user_id is missing or mismatched, or an update policy lacks a with check clause. Reads fail silently by filtering out rows, while writes throw an explicit error. The recommended fix is to write per-operation policies that restrict access to rows where user_id matches auth.uid(), default user_id to auth.uid(), and keep the service_role key off the frontend. Public read access can be allowed only for genuinely public data.

0
ProgrammingDEV Community ·

Helping Well Means First Checking What the Other Person Wants

A piece in DEV Community's social capabilities series argues that a person's movements can hint at their aim, but similar actions may serve different goals. Using a birdhouse-building scene, it shows that helping by passing a tool can ease a small obstacle without taking over the task the person wants to complete. The article draws on research by Warneken and Tomasello on young children supporting others' goals, while noting that such experiments do not reveal a person's long-term wishes. It suggests that clarifying the scope of help, such as asking which part to assist with, and noticing whether the person can continue on their own helps preserve their agency. The author also notes that goals can shift over time and that inferring someone's goal is different from being entitled to decide it for them.

← NewerPage 352 of 3100Older →