How to tell your AI agent lost context: a state-canary in AGENTS.md and CLAUDE.md
A state-canary is a simple observability pattern for AI coding agents: one state line per response helps you catch context drift, lost scope, and…
Read article →Articles
Articles
Notes on Laravel, architecture, performance, and practical software engineering. Based on real project work, written for engineers and teams who build products.
18 articles found
Part 3. A month with an AI journal: finding patterns in sleep, stress, and workouts
How to analyze an AI journal after the first month: transcription fixes, honest reflection with cited sources, Obsidian dashboards, cost, and privacy.
Read article →
Part 2. Hermes Agent + DeepSeek on Ubuntu: a complete Telegram AI journal setup
A step-by-step setup for Hermes Agent and DeepSeek on Ubuntu: a locked-down Telegram bot, local faster-whisper, a Markdown vault, private Git, and a daily…
Read article →
Part 1. How I turned an old gaming laptop into an AI wellbeing journal
How an old Xiaomi Mi Gaming Laptop became a home AI server: Hermes Agent, DeepSeek, Telegram, and Obsidian for a daily wellbeing journal. A…
Read article →
WordPress Bedrock Docker: local dev and bare-PHP production guide
How to run WordPress Bedrock with Docker Compose locally and deploy to Hostinger without Trellis. The two-.env problem, bare-PHP deploy steps, and every sharp…
Read article →
Claude Code setup: a beginner’s guide to the AI coding CLI
Step-by-step guide to installing Claude Code, configuring CLAUDE.md, understanding permissions, and running your first real tasks. Practical, no fluff.
Read article →
CSS works locally but breaks in production: how LiteSpeed UCSS strips your styles
A real debugging story: the theme worked locally but broke in production because LiteSpeed UCSS stripped CSS rules. How to compare what the browser…
Read article →
Claude Code Subagents: Copy-Paste Agents for Safer, Cheaper Workflows
A practical 2026 guide to Claude Code subagents with copy-paste .claude/agents examples for explorer, planner, reviewer, browser testing, model routing, context isolation, and token…
Read article →
AI Agents in the Development Workflow: A Practical Safety Guide for Developers
A practical May 2026 guide to using AI agents in software development safely: Claude Code, Codex, Cursor, Gemini, MCP, permissions, prompts, browser testing, subagents,…
Read article →
When a Leaked Secret in Git Turned Into a Four-Year History Rewrite
A real-world story of leaked API keys, a panicked git filter-branch that rewrote 4 years of history, and the correct incident response we should…
Read article →
Debugging a Production Site After AI Deployment — What the Browser Sees vs. What You Shipped
After deploying a portfolio site built with AI, I found broken mobile menus, bullet-point badges, and content overflow — all caused by server-level CSS…
Read article →
How I Built and Deployed a Custom WordPress Theme with AI Agents in Under 6 Hours
A complete walkthrough of building a zero-JavaScript, multilingual WordPress theme on Bedrock architecture, deployed to Hostinger — with Claude Code AI doing the heavy…
Read article →