mager.co

Always learning. Always eating. Chicago, currently Japan.

Technology All posts →
  1. 01

    SkillOpt: gradient descent for your SKILL.md

    Microsoft's SkillOpt is the first paper to treat agent skill files as trainable parameters — propose an edit, evaluate on held-out examples, accept only on strict improvement. Here's what it found and what it means for teams building with agents.

  2. 02

    OpenHuman — an open-source agent harness that learns who you are

  3. 03

    What Karpathy's CLAUDE.md taught me about my own setup

  4. 04

    Killing OpenClaw for a native Claude Code setup

  5. 05

    gbrain: Migrating My AI Brain From Flat Files to Semantic Memory

  6. 06

    Claude: Anthropic just shipped most of OpenClaw

  7. 07

    Building a tiny local LLM starter for real projects

  8. 08

    The Compounding Agent: Why Hermes Is More Than Just a Pretty TUI

  9. 09

    Claude: How prompt caching actually works

  10. 10

    How I make tokens last longer

Dining Archive →
Life Stories →