KARS: governing agents as untrusted code
An agent’s real instructions are assembled at runtime from model output, tool results, retrieved documents, and MCP servers, none of which you fully control....
An agent’s real instructions are assembled at runtime from model output, tool results, retrieved documents, and MCP servers, none of which you fully control....
Agents are good at writing code. They just don’t know which API shipped last month.
The Afterburner challenge got me thinking: what’s a skill that would be useful across any repo?
Microsoft Build CLI went public on May 11.
Two weeks ago XRM was a 3-hour proof of concept. Now it’s powering a live application.
Requirements document to working app in a weekend.
Most AI coding tools give you one agent. Squad gives you a team.
What if your AI assistant ran entirely on your device but still had access to up-to-date Microsoft docs?
Windows Notepad has been my go-to tool for stripping formatting. Paste anything in, copy it back out, clean plain text. It’s the one app I trust to not add a...
This blog needed some love. Over 200 posts, many without alt text or meta descriptions. Fixing that by hand would take days.