Copilot SDK and agent security validation: governance is now the real developer-tool feature
A practical guide to what GitHub Copilot SDK GA and third-party coding agent security validation GA change for teams adopting AI coding agents.
Writing my thoughts on software, business, and life.
All of my long-form thoughts on programming, business, and more
Vercel Sandbox Drives: Persistent Workspaces Are Becoming Agent Infrastructure
Vercel Sandbox Drives private beta separates disposable execution from durable agent workspace state. Here is what developer teams should design next.
MCP Apps changes agent UX: tool output is becoming a product surface
A practical guide to MCP Apps, Nuxt MCP Toolkit support, and the engineering checklist for turning agent tool results into sandboxed interactive UI.
Next.js 16.2 made agent-ready debugging a framework concern
How AGENTS.md, browser log forwarding, next-browser MCP, Turbopack, and Adapter API turn Next.js 16.2 into an operational signal for AI-assisted teams.
Copilot billing moved to AI Credits: engineering teams now need AI FinOps
GitHub Copilot usage-based billing changes how teams should manage budgets, PR review automation, runners, model choice, and AI usage metrics.
Codex Sites and plugins: internal tools now need a deployment governance layer
OpenAI’s Codex Sites preview, role-specific plugins, and annotations turn agent output into hosted workspace apps, so teams need rules for access, secrets, review, and ownership.
TypeScript 7 Beta and tsgo: Treat the native compiler as a migration project
TypeScript 7.0 Beta introduces a Go-based native compiler preview. Here is how teams should validate tsgo across CI, editors, JSDoc, and build tooling before switching defaults.
VS Code Agents window and BYOK: teams now need an agent operating model
VS Code Agents window, BYOK, remote sessions, and the agent harness shift AI coding from a feature choice into a team workflow design problem.
Chat SDK now includes AI SDK tools: agent features need permission design first
A practical long-form look at Vercel Chat SDK adding AI SDK tools, and why product teams should design approval boundaries, audit logs, and rollback paths before exposing agent actions.
AI Trend Brief #2 (Week 9 Watchlist): What to Track Next Week
A practical watchlist for next week’s AI news cycle, focused on decision quality instead of information overload.