State of agent skills
1 million agent skills in 7 months. Here's what people are actually teaching agents to do.
Why it matters
The skills.sh registry reached 1M skills in seven months—six times faster than GitHub's first million repos. Demand is shifting from technical skills (18% of installs) toward business operations, writing, and agent workflows. This reveals what enterprise teams are actually deploying agents to do, and where the differentiation is moving: from public generic skills to company-specific judgment.
The key facts
9 to knowskills.sh registry: 1 million skills in 7 months (GitHub: 27 months to 1M repos; App Store: 5+ years; npm: 9+ years)
280 million installs recorded
Software engineering: 18% of installs (largest single category); agent workflows: 15%; business operations and writing: ~11% each
Business operations skills perform 74% above average installs per listing; writing/documents 50% above; cloud/infrastructure 42% above
Top 0.04% of skills (375 total) account for 62% of all installs; top 1.2% account for 94%
Nearly half of all skills installed exactly once; no single skill exceeds 1% of total installs
66% of skills are cross-industry; they account for 87.5% of installs and receive 3.6x more installs per listing than industry-specific skills
Agent workflow/automation skills are the largest category among high-volume listings (100K+ installs)
Anthropic introduced Agent Skills in October 2025; Vercel launched skills.sh registry three months later (January 2026)
Go to the source
Vercel Blogvercel.com
Publisher excerpt: In seven months, the registry grew to one million agent skills and recorded nearly 280 million installs.skills.sh A skill gives an AI agent reusable instructions for a particular job. Agents are capable but generic; they can do many jobs, but they don't know how a specific person, team, or company…