Most Agent Memory Systems are Unfinished Databases

In other words, agent memory is just structured data we haven't gotten around to designing a schema for yet

UUIDs are not great for agent use

but not necessarily for the reasons I thought

My Essentials for Understanding Agent Skills

What I needed to get through my head about Skills to actually use them

MCP vs CLI vs Python Tools

When building a remote Python agent, when should you reach for MCP servers, CLI tools, or native Python functions? A practical comparison of trade-offs.

MCP, Skills, and Friends

MCP servers, skills, hooks, and subagents — the tools that extend an agent harness beyond hardcoded tool schemas.

Structured Output

Why structured output is the linchpin — one field determines whether you have a chatbot or an agent.

The Agent Harness

What Claude Code actually is — an agent harness around the Claude API, with an action loop, tools, permissions, and subagents.

Coding Sandboxes

Sandboxing tools for agent code execution, and how to run untrusted code hands-off.

OpenAI's Data Agent