Coding Agents Get a Room: Docker Sandboxes for Claude Code and Codex
A coding agent is a user on your machine. How Docker Sandboxes put Claude Code and Codex in a microVM of their own, so a prompt injection cannot reach your credentials.
Writing
Notes from the field.
On cloud architecture, enterprise integration, AI agent protocols, and what 20+ years of consulting has taught me about building systems that last.
Browse by tag
Jul 23, 2026
Jul 19, 2026
OWASP Goes Agentic: Securing the Skills Your AI Agents Run
The OWASP GenAI Security Project now covers agentic AI, and its new Agentic Skills Top 10 is the first serious risk framework for the skills we install into coding agents. Notes on why it matters and a video worth watching.
Apr 18, 2026
From Hugo to Astro: Rebuilding This Site with Claude Design and Claude Code
Notes on rebuilding this site over a weekend: Hugo to Astro, a visual refresh from Claude Design, implementation through Claude Code, and a rewritten deploy pipeline.
Apr 12, 2026
Framework Desktop as a Local AI Server: Running 70B Models From My Couch
Setting up a Framework Desktop with AMD Ryzen AI Max+ 395 and 128GB RAM as a headless AI inference server, accessible from every Mac in the house via Ollama and Tailscale.
Jan 23, 2026
How LLMs Are Transforming Enterprise Integration Patterns
Exploring how Large Language Models are reshaping application integration, from automated data mapping to intelligent API orchestration