|
|
Hi there, this is your daily ☕️ DevShot.
|
|
|
In today's DevShot:
|
|
💬 Slack launches Slack Code tool 📦 Popular Rust package hijacked to plant malware, and it's linked to North Korea 🤖 Google's AI coding agent leaves its IDE ⚡ Next.js 16.3 cuts dev memory 90% 🔥 SvelteKit 3 reinvents RPCs 🦀 Rust 1.98 adds fast-math float methods Plus: 🎁 6 other news you might like, 🛠️ 5 tutorials, 🧰 6 tools, and 📚 5 papers.
|
|
Want to get the most out of ChatGPT?ChatGPT is a superpower if you know how to use it correctly. Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done. Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
|
💬 Slack launches Slack Code tool
LINK
|
- Slack rolled out Slack Code, a new channel type built for coding agents and the developers supervising them, spun up automatically when you mention a supported agent that decides your request warrants a full coding session.
- Each channel carries the agent's plan, repo and branch, diffs, pull request details, and a live HTML preview, letting people request changes or halt the agent before it archives the channel as a searchable audit log.
- Currently only agents can create these channels, humans can't, and it's live for Claude, Devin, GitHub Copilot, and Vercel, with ChatGPT coming later, free on any plan via an API without MCP or A2A.
|
📦 Popular Rust package hijacked to plant malware, and it's linked to North Korea
LINK
|
- The Rust Security Response Team pulled a malicious `proc-macro1` crate from crates.io after confirming its build script downloaded a hostile payload, along with several sibling crates spreading the same attack.
- The popular `arrayref` crate had been republished to depend on `proc-macro1`, and the author's other packages `internment` and `append-only-vec` were compromised too, likely via stolen credentials rather than malicious intent.
- Developers should check `~/.cargo/registry/cache` for the deleted versions `[email protected]`, `[email protected]`, `[email protected]`, and any `proc-macro1`, `aovine`, `arone`, `aronenao`, or `tinymember` releases.
|
🤖 Google's AI coding agent leaves its IDE
LINK
|
- Google has expanded Antigravity beyond its own desktop app, releasing extensions that bring its AI coding agent into VS Code, Visual Studio, JetBrains IDEs, and Zed, plus availability through eligible Gemini Enterprise subscriptions.
- The extensions open agent conversations in a side panel with inline diffs, plan inspection, and multi-step task delegation; one account works across editors, and the VS Code version ships now while Visual Studio 2026 support remains in preview.
- Sessions inherit organizational IAM policies, VPC Service Controls, and regional boundaries, but shared credit pools are risky since nontrivial tasks burn 150,000-200,000 tokens and per-user or per-team spending controls only arrive later this year.
|
⚡ Next.js 16.3 cuts dev memory 90%
LINK
|
- Next.js 16.3 shipped from Vercel, cutting Turbopack's development memory use by up to 90% via disk caching and memory eviction, both enabled by default across every existing app.
- Vercel's own dashboard dropped from 21.5GB to 2GB, the same disk cache makes repeat CI builds up to 5.5x faster, and server-side rendering rebuilt on native Node.js streams handles ~22% more requests.
- An opt-in Instant Navigations suite adds Partial Prefetching, DevTools called Instant Insights, and a Playwright helper, though static exports break with it and Appwrite advises adopting it one route at a time.
|
🔥 SvelteKit 3 reinvents RPCs
LINK
|
- SvelteKit 3.0's release candidate ships an experimental feature called remote functions, a fresh take on RPCs that lets individual page components pull fresh server data without reloading the whole page.
- Written in JavaScript or TypeScript and compiled into a lightweight client-side wrapper, a remote function fetches from the server inside the component itself, importing a query or mutation directly with no extra routing or boilerplate.
- Shipping experimentally since SvelteKit 2.27, the feature will stand alongside traditional load functions in v3, avoiding the costly page-wide dynamic marking or top-down route loaders that previously sacrificed type safety.
|
🦀 Rust 1.98 adds fast-math float methods
LINK
|
- Rust 1.98.0 adds algebraic floating-point methods for f32 and f64, letting the compiler reorder addition, subtraction, multiplication, division, and remainder using real-number properties, similar to the -ffast-math option found in other programming languages.
- Because floating-point addition is not associative, chaining algebraic_add calls lets the compiler reorder a sum like a + b + c + d and enable broader loop-vectorization, giving non-deterministic results but never undefined behavior.
- Primitive integer types gain a format_into method that writes into a &mut NumBuffer, bypassing dynamic dispatch from write! formatting and performing similarly to itoa, plus a documentation fix guaranteeing ManuallyDrop and Box stay defined behavior.
|
|
2 Free AI Courses. No Credit Card Needed.5,000+ professionals use Skill Leap to get ahead with AI. Right now, two of their best courses are completely free - Claude 101 and the 14-Day AI Boot Camp. Claude 101 covers prompting frameworks, Artifacts, file analysis, and real-world workflows in 19 lessons. The Boot Camp covers ChatGPT, Gemini, Midjourney, and prompt engineering in 16 lessons. Downloadable workbooks. LinkedIn certificate. Zero cost, no credit card, no catch.
📚 Tutorials & Advice
|
> Timing charts: A Blueprint For SMIL animations: Sketch a timing chart mapping when each piece starts and stops before writing SMIL, the JavaScript-free SVG animation language, so overlapping animations stay readable and easy to retime.
|
|
|
🔗 Other News & Articles
-
The August 17 outage, and the work ahead
LINK
-
Test agent changes with LangSmith preview builds
LINK
-
Azure DevOps remote MCP server Reaches GA, without support for Claude, ChatGPT, or Cursor
LINK
-
From all-or-nothing to task-based OAuth consent
LINK
-
Debian just proposed banning AI code. Here’s why it matters for open source developers & maintainers.
LINK
-
DigitalOcean inference router, Now Cache-Aware: Why the Cheapest model Isn't Always the best Deal
LINK
|
|
📄 Papers & Resources
|
> Benchmark scoring rules can flip which AI performs best depending on judging choices alone, with one dataset swinging results by ~41 percentage points, showing leaderboard rankings often reflect the evaluator's setup rather than true ability.
LINK
|
|
> Random workload generator creates realistic, repeatable test cases for scheduling research on complex real-time systems, sparing researchers the manual effort of building their own and making published results easier to trust and reproduce.
LINK
|
|
> Cyber compliance paperwork can be auto-generated for EU Cyber Resilience Act certification, letting smaller companies produce ~70 audit-ready security justifications backed by traceable evidence, cutting the manual effort of proving products are secure.
LINK
|
|
> Automated web testing using the Cypress tool ran a 27-case test suite with fast, stable results, showing that resilient element tags cut the upkeep needed as an app's interface keeps changing.
LINK
|
|
> Agent evaluation contract gives companies one shared, vendor-neutral way to test AI agents across tools like LangChain and CrewAI, exposing what an agent said and did so unreliable or confidently wrong behavior can be caught consistently.
LINK
|
|
|
|
🧰 Latest Tools
| Emergent: builds a full-stack app from a plain-English description, front end through deploy, and you own the code. Start free today |
|
AnySearch: a search API for AI agents that queries multiple trusted sources in parallel, returning filtered, de-duplicated results as structured data.
LINK
|
|
Prefactor: an evaluation layer that scores agent runs in real time, catching quality regressions and drift before they reach production users.
LINK
|
|
ZooData: converts webpages into structured JSON to cut LLM token usage by 75%, with pre-analyzed Amazon and TikTok e-commerce data for AI agents.
LINK
|
|
CartAI: automates checkout across merchant sites through a single API, handling payments, shipping, and order tracking without building per-site integrations.
LINK
|
|
ditto.site: converts any public URL into clean Next.js or Vite code deterministically, preserving components, design tokens, fonts, and hover states via a free API or MCP server.
LINK
|
|
Replay QA: records browser sessions with full devtools access, turning them into detailed, shareable bug reports so AI-built apps get QA coverage without writing test suites.
LINK
|
|
|
|
💬 How did you find today's edition?
We read every reply — just reply to this email and let us know how we can improve!
|
|
|
