Blog ยท Carlos Seidl (EN) https://ceseidl.com.br/en/blog/ News on technology, AI, Claude, .NET and QA. en A TODO that derails agents and Claude locking down IPs https://ceseidl.com.br/en/blog/resumo-2026-11-06/ https://ceseidl.com.br/en/blog/resumo-2026-11-06/ Fri, 06 Nov 2026 08:00:00 -0300 Wednesday to Friday was about keeping agents under control: Claude can now limit where a key works, Cloudflare split out traffic from personal assistants, Google measured what a TODO comment costs a coding agent, and Copilot got a path for moving .NET apps to Windows Arm64. An open 1T Mistral, 128 MB Claude requests and free Blazor https://ceseidl.com.br/en/blog/resumo-2026-10-30/ https://ceseidl.com.br/en/blog/resumo-2026-10-30/ Fri, 30 Oct 2026 08:00:00 -0300 With the GitHub Universe news digested, Thursday and Friday brought more practical things: bigger limits and a cost API for Claude, another free Blazor toolkit, lockfiles for GitHub actions, an open 1-trillion-parameter model and a QA agent that tries to spend less AI on every run. Moonshot launches Kimi K3.1 Pro and promises open weights by November 4 https://ceseidl.com.br/en/blog/moonshot-kimi-k3-1-pro-pesos-abertos-novembro/ https://ceseidl.com.br/en/blog/moonshot-kimi-k3-1-pro-pesos-abertos-novembro/ Fri, 23 Oct 2026 08:00:00 -0300 Kimi's new model reaches the API with a 1-million-token context window and coding scores close to the leading closed models. Copilot in Visual Studio now accepts your own model and fixes vulnerable NuGet packages from the Error List https://ceseidl.com.br/en/blog/copilot-visual-studio-byom-nuget-vulnerabilidades/ https://ceseidl.com.br/en/blog/copilot-visual-studio-byom-nuget-vulnerabilidades/ Fri, 23 Oct 2026 08:00:00 -0300 The September update, published on 10/22, brings Bring Your Own Model in preview, pull request exploration through the Git agent and a shortcut to update dependencies with security flaws. Anthropic donates the Agent Skills standard to the Linux Foundation's Agentic AI Foundation https://ceseidl.com.br/en/blog/anthropic-doa-agent-skills-agentic-ai-foundation/ https://ceseidl.com.br/en/blog/anthropic-doa-agent-skills-agentic-ai-foundation/ Fri, 23 Oct 2026 08:00:00 -0300 After MCP, it is now the turn of skills: the SKILL.md-based format gets a neutral home, with maintainers from several companies. Anthropic launches Claude Duet, a proactive agent that works in the background and asks for approval before acting https://ceseidl.com.br/en/blog/anthropic-claude-duet-agente-proativo-preview/ https://ceseidl.com.br/en/blog/anthropic-claude-duet-agente-proativo-preview/ Thu, 22 Oct 2026 08:00:00 -0300 In research preview for Pro and Max subscribers in the US, Claude Duet runs on its own virtual machine, uses connected apps and hands finished work back for review. Claude API gets version 2026-10-21, with new SDKs, breaking changes and an Apache 2.0 license https://ceseidl.com.br/en/blog/claude-api-versao-2026-10-21-sdks-apache/ https://ceseidl.com.br/en/blog/claude-api-versao-2026-10-21-sdks-apache/ Thu, 22 Oct 2026 08:00:00 -0300 Anthropic released the first new API version since 2023-06-01, and the new major versions of the SDKs and the ant CLI already use it by default. Pylance now accepts Meta's Pyrefly as its type checker in VS Code https://ceseidl.com.br/en/blog/pylance-pyrefly-type-server-protocol-vscode/ https://ceseidl.com.br/en/blog/pylance-pyrefly-type-server-protocol-vscode/ Thu, 22 Oct 2026 08:00:00 -0300 Microsoft announced the Type Server Protocol, an open standard that separates the Python extension from the type checker, and a partnership with Meta to use Pyrefly instead of Pyright. Microsoft.UI.Reactor: an experimental framework for building WinUI apps in C# only, no XAML https://ceseidl.com.br/en/blog/microsoft-ui-reactor-winui-csharp-sem-xaml/ https://ceseidl.com.br/en/blog/microsoft-ui-reactor-winui-csharp-sem-xaml/ Thu, 22 Oct 2026 08:00:00 -0300 Microsoft introduced an open source framework, based on Model-View-Update, where the UI of WinUI apps is described in C# code and rendered with real WinUI controls. Azure DevOps gets agentic Autofix for code and dependency vulnerabilities https://ceseidl.com.br/en/blog/azure-devops-agentic-autofix-codigo-dependencias/ https://ceseidl.com.br/en/blog/azure-devops-agentic-autofix-codigo-dependencias/ Thu, 22 Oct 2026 08:00:00 -0300 Copilot Autofix in GitHub Advanced Security for Azure DevOps now runs on the Copilot coding agent, proposes fixes for vulnerable dependencies and accepts alerts from third-party tools. Recursive releases Flint, an open coding model built to run on a laptop https://ceseidl.com.br/en/blog/recursive-flint-modelo-codigo-local-apache/ https://ceseidl.com.br/en/blog/recursive-flint-modelo-codigo-local-apache/ Thu, 22 Oct 2026 08:00:00 -0300 Richard Socher's startup released Flint, a 40-billion-parameter MoE model (2.5 billion active) under the Apache 2.0 license, which the company says was largely built by an automated research system. PHP fixes two critical remote code execution flaws; the advice is to update now https://ceseidl.com.br/en/blog/php-correcao-critica-execucao-remota-8-5-12/ https://ceseidl.com.br/en/blog/php-correcao-critica-execucao-remota-8-5-12/ Wed, 21 Oct 2026 08:00:00 -0300 On October 21 the PHP Foundation released versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35, which fix two remote code execution vulnerabilities rated CVSS 9.2. Keys of eight eMudhra intermediate CAs were compromised; browsers have already blocked them https://ceseidl.com.br/en/blog/emudhra-ca-intermediarias-comprometidas/ https://ceseidl.com.br/en/blog/emudhra-ca-intermediarias-comprometidas/ Wed, 21 Oct 2026 08:00:00 -0300 Forged certificates could impersonate any site; Firefox, Chrome, Apple and Microsoft have already blocked the affected CAs, but a lot of non-browser software still needs updating. GitHub changes its Git quorum to handle AI agent traffic https://ceseidl.com.br/en/blog/github-git-agent-scale-quorum-por-referencia/ https://ceseidl.com.br/en/blog/github-git-agent-scale-quorum-por-referencia/ Wed, 21 Oct 2026 08:00:00 -0300 Instead of locking the whole repository on every write, Spokes now serializes per reference and gains virtual replicas backed by Azure Blob Storage. NVIDIA releases Kumo Relational, a model that predicts directly on relational databases https://ceseidl.com.br/en/blog/nvidia-kumo-relational-modelo-dados-relacionais/ https://ceseidl.com.br/en/blog/nvidia-kumo-relational-modelo-dados-relacionais/ Wed, 21 Oct 2026 08:00:00 -0300 The model makes predictions such as churn, demand and fraud without training for each task, and ships with an open library and weights on Hugging Face. Survey: swapping the AI model is easy; migrating what you build around it is the hard part https://ceseidl.com.br/en/blog/pesquisa-stacklok-troca-de-modelo-ia-camada-ao-redor/ https://ceseidl.com.br/en/blog/pesquisa-stacklok-troca-de-modelo-ia-camada-ao-redor/ Wed, 21 Oct 2026 08:00:00 -0300 A Stacklok survey of 520 technical leaders shows that integrations, MCP servers and agent workflows weigh more on lock-in than the model itself. Anthropic adds a Harvard physicist to its Long-Term Benefit Trust https://ceseidl.com.br/en/blog/anthropic-long-term-benefit-trust-christopher-stubbs/ https://ceseidl.com.br/en/blog/anthropic-long-term-benefit-trust-christopher-stubbs/ Wed, 21 Oct 2026 08:00:00 -0300 Physicist Christopher Stubbs has joined the Long-Term Benefit Trust, the independent group that helps Anthropic balance commercial growth with its public benefit purpose.