Blog

Thoughts on AI, software development, and engineering.

So you want to query an LLM directly

Minimal, self-contained examples for querying Claude, OpenAI, and Gemini via raw HTTP, provider SDKs, and LangChain in Python, Bash, and TypeScript.

TokenMaxxing

How to cut LLM costs by structuring prompts, sessions, and model routing more efficiently.

Agentic AI Harness

A visual breakdown of the components that make up a modern agentic AI system.

OWASP Scanner-LLM

A project using an LLM to scan for OWASP vulnerabilties

MCP Message Reference

A quick reference describing the messages used in a Model Context Proocol server and client.