ARCHIVES
Everything, by date.
2026
55 POSTS- I Built a Reviewer Skill From 67 Sourced Records, and the First Eval Failed
- Checking the Evidence That the Harness Matters More Than the Model
- Verification Is the Bottleneck Once Agents Write the Code
- A Creative Agent Harness Has to Make Taste Checkable
- Write a Verdict per Idea, Not a Summary
- When Code Is Cheap, Complexity Is the Budget
- AI-Generated UI Fails in a Fixed Set of Ways
- Does This Need AI at All?
- AI Can Help Reverse Engineer a CAN Bus, as One Stage in a Pipeline
- Seven Levels of RAG and the Failure Each One Fixes
- MCP, a CLI or a Direct API Call
- The Order to Build a Skill In
- Most of an Agent's Reliability Lives in the Harness
- An LLM-Compiled Wiki Instead of RAG, for One Person's Notes
- A Loop Needs a Done-Check
- A Second Model Should Review the Plan
- What VRAM Buys You When the Coding Agent Runs Locally
- Agent Memory Is Four Kinds of Files
- Fewer Skills, Better Results
- A Deliberate Handoff Beats Automatic Compaction
- Research Report: C++ Exposed the Limits of AI Coding
- Why I Rejected Kepano's Flat Vault
- Project: Arkive
- Research Report: The 7 Levels of Building ELITE Websites with Claude Code
- Research Report: AI Agents Can Now Build Their Own UI in Real Time (Personalized to You)
- Research Report: Every UI/UX Concept Explained in Under 10 Minutes
- Research Report: How AI Agents & Claude Skills Work (Clearly Explained)
- Research Report: I Found SECRET Features Claude Code Is Hiding (LEAK)
- Research Report: 92% Of AI Users Are On The Wrong Side Of This Trade. (Why This Should Worry You.)
- Research Report: Your Agent Produces at 100x. Your Org Reviews at 3x. That's the Problem.
- Research Report: Tech Recruiter Exposes ATS & Common Resume Mistakes
- Research Report: They solved AI's memory problem!
- Research Report: Full Guide - Build Your Own AI Second Brain with Claude Code
- Research Report: Claude Code Works Better When You Do This
- Research Report: Clief Notes Q and A: April 4th 2026
- Research Report: Coding Interviews Are Completely Different Now (Here's Why)
- Research Report: These 3 Frameworks Make Claude Code Unstoppable
- Research Report: Your Claude Code Terminal Could Look Like This
- Research Report: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI
- Project: ICM Template
- Project: JobReforgerAI
- Research Report: 5 Claude Code Skills I Use Every Single Day
- Research Report: 7 AI Skills That Employers Are Desperately Hiring For Right Now
- Research Report: Your AI Agent Fails 97.5% of Real Work. The Fix Isn't Coding.
- Research Report: The AI Job Market Split in Two
- Research Report: A Markdown File Just Replaced Your Most Expensive Design Meeting
- Research Report: Vibe Coding Is Making Developers Worse
- Research Report: Why Most Tech Professionals Are Terrible Communicators
- Graph RAG Meets ICM: A Strategy for Enterprise Knowledge at Scale
- AI Agent SDKs vs Frameworks: The New Decision Framework
- Research Report: Building a Content Workspace with Claude's Co-Work Mode
- Folder as a Workspace: A Simple Architecture for AI Workflows
- Why Programming Feels Hard Right Now
- Starting LeetCode from Zero
- Project: Building This Site From HTML Portfolio to Hugo Engineering Notebook