Wednesday 12 August 2026 | Join Free | Upgrade

Together with

Hi there, this is your daily ☕️ DevShot.

In today's DevShot:

🔏 Claude now watermarks all AI text

🤖 SpaceXAI launches Grok Bot agent app

⚡ Nvidia releases open model Nemotron 3.5

🖥️ VS Code 1.133 is out

⚙️ .NET 11 Preview 7 speeds up builds

🐧 OpenAI launches ChatGPT app for Linux

Plus: 🎁 8 other news you might like, 🛠️ 4 tutorials, 🧰 6 tools, and 📚 5 papers.

PR-AF ranks #2 of 42 on Martian's Code-Review-Bench: 0.706 recall, roughly 3x more valid findings than commercial tools, at ~10x lower cost per review.

It plans a review strategy per pull request, runs reviewer agents in parallel, and verifies every finding against your source, dropping anything unprovable.

No hallucinated comments.

Apache 2.0, self-hosted, runs on any open or closed model.

Drops into GitHub Actions.

🔏 Claude now watermarks all AI text LINK
  • Anthropic now embeds invisible, machine-readable watermarks into text from Claude models launched since August 2, 2026, adding a hidden signal that persists even after output is copied out of the chatbot elsewhere.
  • Applied at the model level, the mark works across Claude, the API, Claude Code, Cowork, and Tag, plus models served through AWS, Google Cloud, and Microsoft Foundry, and rolls out worldwide regardless of region.
  • A detected mark only signals text may have been processed by Claude, not authored by it, and heavy editing or paraphrasing can erase it; Anthropic will publish detection guidance and C2PA provenance metadata for PNG, JPG, and SVG later.
🤖 SpaceXAI launches Grok Bot agent app LINK
  • xAI has launched Grok Bot, a beta service offering always-on AI agents that run on a cloud PC to complete real-world tasks for users even when their own machine is off.
  • Because each Bot signs into websites and apps as a normal user, it can operate services lacking an API or MCP integration, and multiple Bots can share a conversation to swap context and delegate tasks in parallel.
  • Bots learn by watching a user perform a task once, saving it as a repeatable routine, and retain memory across sessions; access is limited to SuperGrok Heavy, Cursor Ultra ($200/mo), and Cursor Teams Premium subscribers on desktop and iOS.
⚡ Nvidia releases open model Nemotron 3.5 LINK
  • Nvidia released Nemotron 3.5 Lightning, an open 30-billion-parameter mixture-of-experts model whose reasoning nears the larger Nemotron 3 Super, though both trail Google's similarly sized Gemma 4 31B on Artificial Analysis' Intelligence Index.
  • Pitched for speed and customization over raw benchmarks, Lightning delivers up to 4x faster output and, once post-trained by partners like CrowdStrike and CodeRabbit, matched or beat larger proprietary models on specialized tasks.
  • The model is out on Hugging Face, ModelScope, OpenRouter, and build.nvidia.com as a NIM microservice, alongside NeMo Switchyard, a Rust routing library already being adopted by Kong, OpenRouter, and LiteLLM.
🖥️ VS Code 1.133 is out LINK
  • VS Code 1.133 has shipped, letting developers open the Agents window without signing into GitHub through the experimental chat.agentHost.allowSignedOutWhenUsable setting, though it currently only works with Claude, with Copilot and Codex support planned.
  • You can now switch model providers mid-session between Anthropic and Copilot via the model picker, where Anthropic-listed models bill through Anthropic and Copilot models use your subscription, no agent host reconfiguration required.
  • The release also adds sticky scroll for chat prompts that pins the current prompt to the top for easier navigation, plus an integrated browser that auto-refreshes HTML files whenever they change on disk.
⚙️ .NET 11 Preview 7 speeds up builds LINK
  • .NET 11 Preview 7 shipped with MSBuild server and NativeAOT CLI enabled by default, cutting build times and container overhead for cloud workloads while shifting toolchain defaults engineers should verify before deploying to production.
  • The SDK gains new `dotnet test` flags `--timeout` and `--maximum-failed-tests`, traversal project support, and platform-native container publishing defaults, while C# adds labeled break/continue, union patterns, and closed-type constraint exhaustiveness.
  • Web and desktop updates include auto-paused Blazor circuits, OpenAPI 3.2 Server-Sent Events, and Windows Forms dark-mode fixes; this is the last preview before two release candidates precede general availability on November 10, 2026.
🐧 OpenAI launches ChatGPT app for Linux LINK
  • OpenAI has shipped a dedicated ChatGPT app for Linux, previously the only major desktop platform without one, rolling it out worldwide this week as a preview across several major distributions.
  • The app brings ChatGPT, ChatGPT Work, and Codex to Ubuntu 24.04 and 26.04 LTS, Debian 13, and Fedora 43 and 44, with downstream flavors based on those distros expected to work too.
  • OpenAI trails Anthropic here, which launched a Claude desktop app for Linux about a month ago supporting Ubuntu 22.04 or later and Debian 12 or later.

Stop typing what you could say in 10 seconds.

Wispr Flow turns your voice into clean, professional text inside any app. Emails, Slack, client updates — speak once, send without editing. 4x faster than typing.

📚 Tutorials & Advice

> How to Pretty-Print your kubernetes YAML as KYAML and Why You'd want To: Adopt KYAML, a stricter dialect of Kubernetes config files that quotes strings and uses braces, to prevent indentation and type-guessing errors.
> Building monday Com Sidekick Why capable agents need more than just tools: Splitting one overloaded AI assistant into a planner, specialized subagents, and sandboxes made monday.com's Sidekick more reliable because piling on tools quietly degraded accuracy and speed.
> 10 best AI agent skills for senior engineers in 2026: Install reusable skill files that force AI coding agents to follow set workflows like test-driven development, fetch current documentation, and strip out messy generated code.
> Most of your tech debt is free: Refactor only messy code you actively change, since ugly-but-stable files cost nothing while frequently-edited files quietly rack up the real interest.

🔗 Other News & Articles

  • Taking AUTO CDC to the next level: Solving the hardest real-world use cases LINK
  • Open-sourcing metals v2: Databricks’ Java and Scala language server for multi‑million line codebases LINK
  • Microsoft releases MAI-Code-1.1-Flash coding model to better compete with Chinese models LINK
  • MCP goes stateless, and developers Ask whether that Just makes It an API Again LINK
  • Electric joins Databricks to bring WASM Postgres to AI agent sandboxes LINK
  • CNCF Announces graduation of cloud native buildpacks, Advancing the standard for container builds LINK
  • SQL server management studio 22.9 brings modern connection dialog updates and lots more LINK
  • Software engineering at a proprietary trading company: Optiver LINK

📄 Papers & Resources

> Software fingerprint matching gets more reliable when compilers merge functions together, helping security teams spot copied, vulnerable, or malicious code across programs that current tools misread after this common optimization. LINK
> Self-improving coding assistants that rewrite their own instructions across eight programming languages mostly boost problem-solving rates and share a common core strategy, but each language still needs its own custom tuning to fully work. LINK
> Smart contract bug scanners flag many false alarms, but adding rule-based checks to Mythril's reports cuts those false positives, letting teams trust vulnerability alerts and fix real security flaws faster. LINK
> Theory-to-code translation tested whether formatting a theory as structured rules versus plain prose changes the computer model an AI builds from it, finding it mostly does not, since only 19 of 108 checks showed a difference. LINK
> Automated bug fixing feeds AI coding tools only the code that actually caused a crash, instead of noisy leftover context, fixing 313 bugs on a standard benchmark for about 3 cents each. LINK

🧰 Latest Tools

Sim: a visual and code-based workspace for building and deploying AI agents, connecting to 1,000+ integrations and every major LLM provider. LINK
Prelint: reviews AI-generated pull requests against your ADRs, docs, and past decisions to catch product drift before it ships. LINK
SKI: lets you voice-code with Claude Code, Codex, and other agents, getting spoken replies back so you build hands-free at thinking speed. LINK
Humalike x Hermes: provides turn-taking, timing, and memory APIs so AI agents know when to speak, wait, or interrupt naturally in conversation. LINK
Coldtea.ai: automates QA, monitoring, and bug detection with AI agents that flag issues and create fix tasks post-deployment. LINK
Openbase: helps developers pick reliable open-source packages by comparing popularity, activity, and reliability metrics alongside real user reviews. 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!

★★★★★  Nailed it
★★★  Average
  Fail

Not subscribed to ☕️ Devshot yet?Subscribe for free