Journal
Posts and long-form series on AI, startups, venture capital, and more.
All Posts
Ensemble Rating Systems: Combining the Best of All Worlds
Combine multiple rating algorithms like Elo, Glicko, and TrueSkill to get better predictions. Learn how Elote's ensemble approach works.
uvx --from: Run Any Python Tool Directly from GitHub
You can use uvx to run Python CLI tools straight from a GitHub repo with no install step. Here's how it works and why it's useful.
Building idea.log with SwiftUI and SwiftData
Notes on shipping a SwiftUI app with SwiftData persistence, App Intents for Siri, semantic search, and a feature that exports ideas as structured AI prompts.
Sentinel: Using MCP Sampling to Analyze Your Own Tool Security
Building an MCP server that uses sampling to discover your connected tools and analyze them for security risks, with no API keys required.
The Colley Matrix: A Linear Algebra Approach to Ratings
The Colley Matrix uses linear algebra to solve for optimal ratings simultaneously. Originally designed for college football rankings.
MCP Sampling and Elicitation: The Features That Make Servers Smart
A technical guide to MCP sampling and elicitation: two powerful features that let MCP servers talk back to the client's LLM and ask users for input.
A Practical Guide to GitHub Branch Protection Rules
What each GitHub branch protection rule actually does, when to use it, and how to choose the right configuration for your project's size and risk profile.
Stargazers Now on Homebrew
Stargazers, the CLI for analyzing your GitHub audience, is now installable via Homebrew along with new features for account-wide trend analysis.
The DWZ Rating System: German Engineering for Chess Ratings
The Deutsche Wertungszahl (DWZ) chess rating system combines German precision with practical considerations. See how it works with Elote.
makefile-mcp: Output Caching, Pagination, and Search
makefile-mcp now caches full execution output server-side, with tools to paginate and search through results from previous runs.