Vercel's skills.sh registry hits one million agent skills in seven months

What a skill is and why it spreads quickly

A skill gives an AI agent reusable instructions for a particular job. Vercel's report argues that agents are capable but generic: they can do many jobs, but they do not know how a specific person, team or company does them. A skill supplies that missing context and can be as simple as a file written in ordinary language.

Anthropic introduced Agent Skills in October 2025, and Vercel launched the skills.sh registry three months later. Writing a skill means making some of the judgment behind a job explicit, such as the steps to follow, what good looks like and how to evaluate the result. Because the agent and its tools already provide much of the underlying capability, Vercel says skills are faster and easier to create than conventional software.

The company compares the pace with other platforms. According to the report, skills.sh reached one million skills seven months after launch, against 27 months for GitHub repositories, 63 for App Store apps and 117 for npm packages. Vercel's explanation is that far more people can describe how work should be done than can program a computer to do it, and a finished skill can be installed without being rewritten for each agent.

Supply is technical, demand is broader

Vercel classified the registry's most-installed skills by the kind of work they cover. These skills account for more than four-fifths of all installs. On the supply side, more than half of listings teach software engineering, agent workflows, data, infrastructure or security, with software engineering alone making up about a quarter.

Demand looks different. No category holds more than a fifth of installs. Software engineering is still the largest at 18%, followed by agent workflows at 15%, and business operations and writing at nearly 11% each.

Comparing installs per listing shows where supply and demand diverge. The average business operations skill draws 74% more installs than the average, writing and documents 50% more, and cloud and infrastructure 42% more. Software engineering skills draw about 30% fewer installs than average, education and productivity 39% fewer, and research 55% fewer. Vercel concludes that most skills are technical because they began in developer tools, while installs cover a broader range because agents are useful across a company.

• Agent workflows and automation account for 14.8% of installs in the classified dataset.
• Among skills with at least 100,000 installs, agent workflows are the largest category.

Vercel notes that agents also help write skills, so people are using agents to improve agents.

Installs are concentrated, but there is no single winner

Install activity is highly concentrated. Nearly half of all skills were installed exactly once. At the other end, 375 skills, or 0.04% of the registry, account for 62% of installs, and the top 1.2% account for 94%. Even so, the most-installed skill accounts for less than 1% of all installs.

The report explains this by saying skills compete within jobs and accumulate across them. A team that picks one expense-report skill has little reason to add another, but it can use that skill next to others for spreadsheets, research and presentations.

Cross-industry skills, such as cleaning up a spreadsheet or deploying a website, make up 66% of classified skills and 87.5% of installs. They receive 3.6 times as many installs per listing as industry-specific skills. Only one in eight installs is tied to a specific industry.

What comes next

Vercel expects the next phase to move toward company-specific knowledge. In its words, "The first million skills taught agents what everyone knows. The next million will teach your agents what only you know." The company argues that public skills and better models will make general expertise a baseline, so more differentiating value will sit in judgment such as when a customer gets a refund or what can ship without another review.

It also expects skills to be judged on effectiveness rather than popularity. Today, install count is the best signal of quality, but Vercel says a skill justifies itself only if the agent does the job better with it than without it, and predicts tests and benchmarks that measure exactly that. Organizations, it says, will write down the knowledge only they have and maintain it the way they maintain code.

Vercel cautions that install figures use aggregate registry counters and do not represent unique people or necessarily independent choices. Function and industry findings describe a classified sample rather than the full catalog, and figures may be revised.