What Is AI-DLC? Understanding AWS's Proposed Approach Through Examples
An explanation of AWS's AI-Driven Development Life Cycle through its current five phases, official examples, enterprise case, and adoption checklist.
Harness Engineering
24 posts with the tag “Harness Engineering”
Designing Self-Healing Loops for Coding Agents
A practical guide to coding-agent self-healing loops based on 2026 research and official implementations from OpenAI, Google, and GitHub, covering layered verification, stop conditions, permission boundaries, and human approval.
How AI Development Rules Protect Color Accessibility
How I am building a harness to reduce repeated color accessibility issues in UI components created through Vibe Coding, especially across light/dark themes, hover, selected, and focus states.
Codex Memories: What to Store and When to Reset
A practical guide to what Codex Memories are for, how they differ from AGENTS.md, and when to avoid, review, or reset memory behavior.
Claude Code to Codex Handoffs: Three Ways to Preserve Context
How I decide when to use Claude Code versus Codex in the same repository, and how I design handoff documents to transfer context across sessions.
How prebuild Normalization Prevents Validation-Order Failures
How I reduced recurring Astro build failures by moving automatic normalization into prebuild instead of treating each failure as a one-off fix.
How to Verify AI-Implemented UIs with Playwright
How AI Learning Playground positions Playwright alongside structural checks, URL verification, and human review.
Three Months of Vibe Coding: From Development to Maintenance
A look back at three months of building and running AI Learning Playground with Vibe Coding, organized around insights from the technical, design, and operational phases.
Claude Code Hooks: Setup and Automated Check Examples
A plain-language guide to using Claude Code Hooks for repeated checks, including before-versus-after timing, human approval boundaries, and an adoption process.
How to Preserve AI Workflow Decisions in lessons.md
How to record an AI workflow problem, cause, resolution, and prevention in lessons.md as a decision history that later work can verify.
Approval Gates for AI Agent Build Commands
A record of how a production build command ran while I was asking AI to make a separate change, and the approval-required rule I put in place afterward.
How to Hand Off a Dirty Worktree Between AI Tools
A safe handoff process for uncommitted changes, separating what Codex Local, Worktree, and Cloud environments can see.
How to Validate Internal Links Against Public URLs
A file exists in the repository but the link is broken — this happened on this site. Here is why verifying internal links by file path alone is insufficient, and how I addressed it.
How to Prevent Unrequested UI Changes in AI Development
While asking AI to add a blog article, I discovered that the navigation layout had also been changed without instruction. This article describes why that happened and how I addressed it with a rule in CLAUDE.md.
Building Validation Scripts with AI to Automate Manual Checks
Manually checking the consistency of configuration files has limits. This article covers the process of building a validation script with AI and why automated detection proved useful.
Automatic learning_time Normalization: Design and Implementation
I updated the learning_time workflow so docs articles normalize reading time during authoring, while dev:check and prebuild keep the same normalization as a safety net.
How to Prevent JA-EN Blog Drift with Structural Checks
This article explains the current workflow for synchronizing English articles from canonical Japanese, using three translation-drift patterns and a two-stage checklist. It separates differences that structural checks can surface from semantic differences that require human judgment.
What Is SKILL.md? Standardizing Repeatable AI Work
Repeating the same instructions to an AI every time is less stable than defining procedures as a skill file. This article covers how I designed SKILL.md files for this site and what changed as a result.
What Is Spec First? Reducing Rework in AI Implementation
Writing a specification before asking AI to implement reduces mismatches and repeated revisions. This article explains the Spec First principle and what to include in a simple specification.
Sharing Claude Code and Codex Rules in shared/
When the number of rule files for AI grows, knowing what is where becomes difficult. This article explains how I organized them into three categories—rules, skills, and workflows—using the shared/ directory.