Insights
Founder and builder perspectives on AI tools, thinking patterns, and the new way of working
Showing 229-240 of 336
Banning Inter: Why Font Defaults Are the Slop Tell
Inter, Roboto, Arial, Space Grotesk — the fonts models default to and why they read as AI-generated. What committed type pairing looks like instead.
Don't Install 45 Skills. Stack These Four.
You don't need a pile of design skills to beat AI slop. You need four: a taste base, motion, tokens, and a visual feedback loop. Here's the recipe.
The Six-Layer Anti-Slop Stack
AI design slop lives on six layers, from frontend UI to trust and orchestration. Here's the full map and the skill that fixes each one.
What a Design Skill Actually Is (and Why It Beats a Prompt)
A design skill is a plain-markdown SKILL.md the agent loads on every turn. Here's why that beats a one-off prompt, plus the three ways to install one.
AI Design All Looks the Same. Here's the Tell.
Every vibe-coded site has the same signature: purple gradient, Inter, four cards, one weak hover. Here's why models ship it and how to overwrite it.
Prompt Engineering Is Out. Loops Are In.
Why loop engineering replaced prompt engineering in 2026, and what changes for builders when the unit of work becomes an autonomous loop instead of a prompt.
The 'Build it from Scratch' Discipline in the Age of Skills
Karpathy's pedagogy prizes building from first principles. The Claude Code skills ecosystem prizes install-and-ship. These aren't in conflict — but the tension is real and worth naming.
CLAUDE.md as Context Engineering: Why Karpathy's Practice Matters
CLAUDE.md files have become the standard context-injection point for Claude Code projects. Karpathy's coding guidelines show exactly why the practice is more powerful than it looks.
From nanoGPT to Claude Skills: The Pedagogy of Karpathy's Public Code
Karpathy's 'build it from scratch' lineage — nanoGPT, micrograd, makemore — encodes a theory of understanding that ports directly into how the best Claude skills are designed.
Karpathy on Agents: His Public Takes on Autonomous Coding
Karpathy named December 2025 as the inflection point for agentic coding. Here's his actual framework for thinking about agents, the limits he flags, and what it means for Claude Code workflows.
The Karpathy Skills Repo: A One-Skill Field Guide
The multica-ai/andrej-karpathy-skills repo contains exactly one skill. That single-skill design is itself a statement about how to write useful AI guidance.
Reading Karpathy's Guidelines: What His One Skill Reveals
A close reading of the karpathy-guidelines skill — four rules, 65 lines, no ceremony. What each rule actually does and why the structure matters.