AI agents
An Agent Sandbox Is Useful Only When It Has Real Boundaries
GitHub's new local sandboxing preview is a useful reminder that an approval dialog and an enforceable boundary solve different problems.
The engineering journal
01
02
03
The engineering journal
The engineering journal
Behind the build.
Beyond the demo.
Experiments, practical guides and lessons from building AI systems. The decisions, the details, and what I learned along the way.
Featured · AI agents
GitHub's new local sandboxing preview is a useful reminder that an approval dialog and an enforceable boundary solve different problems.
The library
6 articles
AI agents
GitHub's new local sandboxing preview is a useful reminder that an approval dialog and an enforceable boundary solve different problems.
ai
Learn how to set up CLIProxyAPI to use GPT, Claude, and Gemini models inside Claude Code on Windows, with separate profiles, automatic launchers, and proper local configuration.
ai
The uncomfortable truth about using AI for 12 months without understanding its cognitive process and the exact framework that puts you in the top 0.01% of users.
ai
Learn how to integrate powerful AI capabilities into your Next.js applications using Groq's lightning-fast inference engine and the Vercel AI SDK.
ai
A comprehensive guide to prompt engineering, covering techniques from basic to advanced that will help you get the most out of large language models.
career
Reflecting on my transition from frontend development to becoming a full stack AI engineer, the challenges I faced, and the lessons I learned along the way.
New articles on AI systems, research and building software. Confirm your email to subscribe, or follow the RSS feed.