Writings

When an agent fails, harness engineering gives you four levers (Context, Tools, Loop, Governance) to find which one broke in under a minute.

Prompt-based approval gates fail because the model decides whether to ask. Mastra's requireApproval primitive removes that decision entirely. Here's how to implement it.

What 'supported by a fleet of agents' means in practice: which tasks automate, which don't, and where the ROI breaks down. Evidence from Stripe, Coinbase, Ramp, and Shopify.

A backend PR kept colliding with other merges on database evolution numbers. Four manual rebases later, I described the problem to Claude and let it write a routine to handle the rest.

Practical tips for getting the most from Claude Opus 4.7's 1M context window in Claude Code. Effort levels, proactive compaction, subagent delegation, and session management from daily production use.

Most AI agent governance advice targets boards, not builders. Three failure patterns, real TypeScript examples, and what a CTO should do Monday morning.

How I added ElevenLabs TTS audio narration to my Hugo blog, cloned my own voice, and discovered my writing had patterns no voice model could read.

How I used autoresearch to run 65 autonomous prompt optimization iterations on a production LLM agent, cutting it 28% while retaining 98% output quality.

AI agents produce better output when the codebase is ready for them. Here are the four dimensions of codebase readiness that account for most of the gap.

Claude Code runs terminal commands and asks you to approve them. This explains what those commands mean and when to pause before saying yes.