Blog
AI for Developers
I cover practical ways to use AI in daily dev work: debugging stack traces, generating unit tests, refactoring large classes, and reading PostgreSQL query plans. For developers who want AI to save real hours without trusting it blindly.
Claude API 400 Error: temperature, top_p, top_k
Claude API 2026-10-21 returns a 400 if you send temperature, top_p or top_k. See what breaks and how to fix it.
AI Prompt to Turn a Bug Report Into a Test
An AI prompt that turns a vague bug report into a reproducible failing test, so you fix the real cause instead of guessing and hoping it sticks.
Reduce Claude Code Token Usage: 9 Ways
How to reduce Claude Code token usage: where tokens go, how to track them, and nine practical ways to use fewer.
Claude Context Window Size by Model (2026)
Claude context window size for every current model across the API, app and Claude Code, and what happens when it fills.
AI Refactoring: Breaking Up a God Class
A real AI refactoring workflow for breaking up a legacy god class safely: the prompts that work, and the ones that quietly break production.
Fix Invalid JSON from LLM API Responses
Why LLM APIs return invalid JSON and how to fix it for good: structured output modes, schema validation, and a safe parse-and-repair fallback in Node.js.