Insights

Top 15 Claude Code skills for testing and TDD. The #1 and #2 entries aren't test runners — they're TDD methodology skills at 4/5 quality, totaling 290K install signal.

Top 15 Claude skills for video production. Remotion dominates (4 entries); pixel-art at #1 is a tag-miscategorization with high Hermes signal.

Every official Anthropic-published Claude skill, ranked. All 11 carry a 5/5 quality score; Frontend Design is 57× more installed than the next entry.

Fifteen skills for building, running, and improving AI agents — ranked across 465 candidates. The dominant 2026 pattern: agent memory and self-improvement. 10 of the top 15 are about either remembering or evolving.

Fifteen Claude Code skills ranked for the day-to-day developer workflow — code review, debugging, git, testing, auth, frontend. obra/superpowers takes the top 3.

Twelve skills for the operational side of AI-assisted development — auto-updaters, MCP management, Docker, git, tmux. The top picks are about keeping your AI stack healthy, not deploying it.

Twelve skills for processing, transforming, and humanizing documents. The most installed: Summarize, at 171K signal. The most uncomfortable trend: three of the top 9 are AI-text humanizers.

Fifteen productivity skills ranked from 454 candidates. The category splits cleanly: meta-cognition (plan/execute) versus office-app integrations (Notion, PowerPoint, Lark). Both win.

Twelve security skills ranked. The most surprising finding: the top of the list isn't about securing your code — it's about securing the skills you install before they touch your code.

RAG-based code search hits a wall at scale: the index is stale before it ships. Claude Code's grep-and-read approach trades upfront indexing for live traversal — and that tradeoff scales differently than most teams realize.

Anthropic's Founder's Playbook quietly settled an argument I've been having with founders for a year: which Claude surface do you use, when, and why. Here's the decoder, and what each one is structurally good and bad at.

Instructions that helped your old model can hurt your new one. Anthropic's Applied AI team just put a number on how often you should audit your CLAUDE.md, hooks, and skills — and gave two concrete examples of rules that aged badly.