 |
VOOZH | about |
|
VOOZH | about |
We’re so glad you’re here. You can expect all the best TNS content to arrive Monday through Friday to keep you on top of the news and at the top of your game.
Check your inbox for a confirmation email where you can adjust your preferences and even join additional groups.
Follow TNS on your favorite social media networks.
Become a TNS follower on LinkedIn.
Check out the latest featured and trending stories while you wait for your first TNS newsletter.
In October, Anthropic launched Agent Skills as a way to teach its Claude service repeatable workflows. Building those skills involved, for the most part, using Claude itself and the interactive “skill-creator” skill, which generated the folder structure and a SKILL.md file. Today, Anthropic is launching the Agent Skills spec as an open standard in a move that feels similar to how it turned its Model Context Protocol (MCP) into the de facto standard for how AI agents use tools today.
“With the open Agent Skills standard, we’re radically expanding the ability to create and deploy shareable, simple-to-implement, powerful, and portable skills — accessible by all,” an Anthropic spokesperson told me. “This unlocks a new era of AI agents that are built to power real-world enterprise tasks.”
The SKILL.md file is a mix of YAML and Markdown content that defines the skill name and includes step-by-step instructions for Claude to follow, as well as examples of how the skill is used.
Skills are available in the Claude apps (once enabled in the settings) and in Claude Code once installed in the plugin directory. Claude API users can access them via the /v1/skills endpoint.
Over time, Anthropic surely hopes that the Agent Skills spec will become the standard for defining these workflows. “We’ve been collaborating with members of the ecosystem, and we’re encouraged to see early adoption of the standard,” the company writes in today’s announcement.
In addition, the company is also launching a directory with skills from its commercial partners, including the likes of Atlassian, Canva, Cloudflare, Figma, Notion, Ramp and Sentry.
As before, Claude users will still be able to build their own skills easily. Like before, the way to create a skill is as easy as describing what it should do. One new feature is that Claude will now show a preview of the full contents of the skill package.
Skills will now also get a more prominent spot in the Claude interface.
To some degree, skills were always meant to appeal to enterprise users, but until now, Anthropic didn’t offer IT admins many controls over how they could be used. That’s also changing today, with the launch of central admin settings that will allow admins to set which skills are provisioned for their users and which ones are enabled by default.
This new feature will be available to Team and Enterprise administrators.
Anthropic’s new skills org management interface. Credit: Anthropic.
“Atlassian’s skills bring our decades of teamwork expertise and best practices to Claude,” explained Josh Devenny, the head of product for Rovo Skills at Atlassian. “Now Claude doesn’t just see Jira tickets or Confluence pages; it knows what to do: turning specs into backlogs, generating status reports, surfacing company knowledge, triaging issues, and more.”
We’re currently seeing efforts from a number of the frontier model providers to build more of a skills or app ecosystem on top of their core models. OpenAI tried this with the GPT Store last year and now again with apps that extend ChatGPT. Google offers Gems for building custom, specialized agents and the Microsoft Copilot ecosystem features declarative agents.
Discoverability will likely remain a problem for these tools in the short term, though. The AI tools keep changing rapidly, after all, and most casual users can barely keep up with the feature launches. For Anthropic, the skills directory will surely help there, but new users need to know it exists in the first place.
