Can personal context servers improve AI outputs?
A blind-judged test suite pits my personal MCP server against unaugmented LLMs, ruling out confirmation bias along the way.
A blind-judged test suite pits my personal MCP server against unaugmented LLMs, ruling out confirmation bias along the way.
Two small Python scripts for carrying your Bluesky mute and block lists across to Mastodon.
I had a website that felt like me, then I didn't. This is the attempt to put it back.
How I encoded fifteen years of design leadership thinking into a portable knowledge base that AI can retrieve and reason from.
No notes match.