What are Claude Code skills exactly?

Markdown files in .claude/skills/ with name, description, body. Model tool-calls to load full instructions.

Should I use CLAUDE.md for TDD guidelines?

Yes—always-in-context beats invocation roulette. Skills for one-offs only.

Why did Vercel's evals favor AGENTS.md?

Single-shot tests; ignores multi-turn accumulation where skills could thrive if called.

How does Superpowers bypass this?

Pre-prompt injection, faking CLAUDE.md delivery without tool dependency.

☁️ Cloud & Infrastructure

Claude Code Skills Flop 94% of the Time: Ditch Them for Best Practices

Everyone thought Claude Code's skills would bundle best practices smoothly. Turns out, the model ignores them most of the time—exposing a brutal activation gap.

theAIcatchup Apr 10, 2026 3 min read

Bar chart showing Claude Code skill invocation rates vs CLAUDE.md baselines in multi-turn evals

⚡ Key Takeaways

Skills match CLAUDE.md quality but invoke only 6-66% of the time— that's the real killer. 𝕏
Put best practices in always-on CLAUDE.md; reserve skills for on-demand recipes. 𝕏
Superpowers succeeds by hacking around the invocation gap, not fixing it. 𝕏

Published by

theAIcatchup

Ship faster. Build smarter.

#AGENTS.md #AI skills #CLAUDE.md #Claude Code

Worth sharing?

Get the best Developer Tools stories of the week in your inbox — no noise, no spam.

Originally reported by dev.to

⚡ Key Takeaways

The 60-Second TL;DR

theAIcatchup

Share this article

Worth sharing?

Related Stories

Subdirectory CLAUDE.md Files: The Brain Hack That Supercharges Claude Code Agents

Claude Code's Five-Layer Memory: Smart Forgetting in AI Coding

Claude's Secret Emotions: Guardrails for Coding Agents

I Slashed Claude Code Token Waste by 65%—Here's the File Structure Secret

Stay in the loop