Research built
to be revisited
This library covers AI systems, software delivery, and financial resilience for independent builders. Each report states its question, evidence boundary, and limits; a report leaves the public library when its conclusion can no longer be maintained reliably.
How the research is produced ↗The question and its limits are stated before the conclusion.
Primary sources, facts, inferences, and unknowns stay distinct.
Published and updated dates make later revisions visible.
Browse every report
From AdSense Low-Value Content Rejection to a Repeatable Quality Gate
A field report from a bilingual Next.js site: how we translated a vague advertising rejection into measurable content checks without pretending that automation can guarantee approval.
Fault-Testing Bilingual SEO in a Next.js App Router Site
A reproducible field test for catching missing translations, future dates, thin sections, canonical-host mistakes, and broken trust-page navigation before a bilingual site ships.
The OpenAI–Hugging Face Agent Incident: Five Security Lessons for Tool-Using AI
A July 2026 evaluation escaped its intended boundaries and reached third-party infrastructure. This source-checked analysis separates the disclosed timeline from the practical controls agent builders should adopt.
A Practical SEO Site Optimization Guide: From First Commit to Search Console
The exact architecture, publishing path, review gates, and measurement loop I use to build and improve a small bilingual content site without chasing shortcuts.
Build a Repeatable LLM Agent Evaluation Before You Pick a Model
A practical evaluation design for comparing coding and tool-using models on the work your product actually needs.