Anthropic said on Wednesday that he would release it Agent skills The technology, as an open standard, is a strategic bet that sharing its vision to make AI assistants more capable will solidify the company’s position in the rapidly evolving enterprise software market.
The San Francisco-based artificial intelligence company also added organization-level management tools for enterprise users and a directory of partner-built skills from companies. Atlassianfor , for , for , . Figmafor , for , for , . Canvafor , for , for , . stripfor , for , for , . ideaand Zippier.
The move marks a significant expansion of a technology Anthropic first introduced in October, turning what started out as a niche developer feature into an infrastructure that’s now poised to become an industry standard.
"We are launching Agent Expertise as an independent open stander with a specification and reference SDK available. https://agentskills.iofor , for , for , ." Mahesh Murg, Product Manager at Anthropic, said in an interview to VentureBeat. "Microsoft has already adopted Agent expertise within Versus Code and GitHub. So there are popular coding agents like Cursor, Goose, AMP, OpenCode, and more. We are in active dialogue with others in the ecosystem."
Inside the technology that trains AI assistants to perform specialized tasks
Skills, at their core, are folders containing instructions, scripts, and resources that tell the AI ​​system how to perform specific tasks consistently. Every time users need an AI assistant to complete a specific task, a skill package needs to be provided to an AI assistant in a reusable module.
This concept addresses the main limitation of major language models: although they possess extensive general knowledge, they often lack the specific procedural skills required for specialized professional work. For example, a skill for creating PowerPoint presentations might include preferred formatting conventions, slide structure guidelines, and quality standards.
Anthropic designed the system around this "Progressive disclosure." Each skill takes only a few dozen tokens when summarized in the AI’s context window, with full details only when the task requires it. This architectural choice allows organizations to deploy extensive skill libraries without overwhelming the AI’s working memory.
Fortune 500 companies are already using skills in legal, finance and accounting
Administrators in the new Enterprise Management have Anthropic Administrators permissions The team And Enterprise Centrally skill delivery projects, allowing individual employees to customize their experience, controlling which workflows are available in their organizations.
"Enterprise users are using the expertise in production, both in coding workflows and in business functions such as legal, finance, accounting, and data science." Death said. "Feedback has been positive as the skills they give Claude personalize how they actually work and quickly achieve high-quality production."
According to Morag, the response from the community has exceeded expectations: "Our skills repository already surpassed 20K stars on GitHub, with tens of thousands of community-created and shared skills."
Atlassian, Figma, Stripe, and Zapier join Anthropic’s expertise directory at launch.
Anthropic is launching with the expertise of ten partners, a roster that reads like modern enterprise software. The presence of Atlassianwhich makes Cumin And The confluencealong with design tools Figma And Canvaa payment infrastructure company stripand automation platforms Zippier suggests that expertise is positioning as the connective tissue between the anthropic cloud and the business of applications.
Business arrangements with these partners are focused on ecosystem development rather than immediate revenue generation.
"Partners who develop skills for the directory do so in order to improve how Cloud works with its platform. It is a mutually beneficial ecosystem relationship similar to the MCP Connector Partnership," Murg explained. "There are currently no revenue sharing arrangements."
To test new partners, Anthropic is taking a measurement approach. "We started with established partners and are developing more formal standards as we expand," Death said. "We want to create a valuable supply of expertise for businesses, helping partner products shine."
Specifically, not charging extra for anthropic capacity. "Skills work at the layers of Cloud: Cloud.E, CloudCode, CloudAgent SDK, and API. They are included in the Max, Pro, Team, and Enterprise plans at no additional cost. API usage determines standard API pricing," Death said.
Why Anthropic Is Giving Its Competitive Advantage to OpenAI and Google
Release decision Expertise as an open standard A calculated strategic choice. By making expertise portable across AI platforms, Anthropic is betting that ecosystem growth will benefit the company more than proprietary lock-in.
The strategy is working. Openei has quietly adopted a structurally similar architecture in both Chat GPT And his Codex CLI Tool. Developer Elias Juden discovered the implementation earlier this month, looking for directories containing skill files that mirror Anthropic’s specification. Same file naming conventions, same metadata format, same directory organization.
This gathering suggests that the industry has found a common answer to a vexing question: How do you make an AI assistant consistently good at a specialized task without expensive model fine-tuning?
The timing coincides with broader standardization efforts in the AI ​​industry. Anthropic Donated Its Model Context Protocol was presented to the Linux Foundation on December 9, and both Anthropic and OpenAI co-founded Agentic AI Foundation Along the block. Google, Microsoft, and Amazon Web Services joined as members. The Foundation will offer a number of open specifications, and specialization will fit naturally into this standardization push.
"We have also seen how complementary skills and MCP servers are," Murg noted. "MCP provides secure connections to external software and data, while expertise provides procedural knowledge to use these tools effectively. Partners who have invested in MCP’s strong integration were a natural starting point."
The AI ​​industry abandons special agents in favor of an assistant that learns everything
Skill The approach is a philosophical shift in how the AI ​​industry thinks about making AI assistants more capable. The traditional approach involves building specialized agents for different use cases. A customer service agent, a coding agent, a research agent. Expertise suggests a different model: a general-purpose agent equipped with a library of specialized abilities.
"We used to think that agents in different domains would look very different," According to one tireless researcher, Barry Zhang, told an industry conference last month Business Insider Report. "The agent below is actually more universal than we thought."
This insight has important implications for enterprise software development. Instead of building and maintaining multiple specialized AI systems, organizations can invest in creating and curating skills that encode their institutional knowledge and best practices.
Anthropic’s own Internal research supports this view. A study the company published in early December found that its engineers used the cloud for 60 percent of their work, boosting self-reported productivity by 50 percent, a two- to three-fold increase from the previous year. Notably, 27% of the tasks Claude helped involved tasks that wouldn’t otherwise be done, including creating internal tools, creating documents, and addressing what employees called. "Paper cut" – Improvements in small standards of living that were permanently eliminated.
Security risks and skill atrophy emerge as concerns for enterprise AI deployments
Skills framework Not without potential complications. As AI systems become more capable through expertise, questions arise about maintaining human expertise. Anthropic’s internal research found that while skills have enabled engineers to work in more domains — backend developers build user interfaces, researchers create data visualizations — some employees worry about skills atrophy.
"When producing output is so easy and fast, it becomes harder and harder to take the time to actually learn something." said an Anthropic engineer in an internal survey of the company.
There are also security considerations. The skill provides the cloud with new capabilities via instructions and code, which means a malicious skill could theoretically introduce vulnerabilities. Anthropic recommends only installing skills from trusted sources and thoroughly auditing those from less reliable sources.
Open standard The approach also introduces questions of governance. While Anthropic has published the specification and launched a reference SDK, long-term responsibility for the standard is unclear. Whether it will fall under the Agentic AI Foundation or need its own governance structure. This is an open question.
Entropic’s real product may not be the cloud—it may be the infrastructure that everyone else builds
The speed of mastery reveals something important about Anthropic’s ambitions. Two months ago, the company introduced a feature that looked like a developer tool. Today, this feature has become a specification that Microsoft has built into VS Code, which OpenAI replicates in GPT, and the enterprise software giants are racing to support it.
This pattern echoes strategies that have reshaped the technology industry before. Companies from Red Hat to Google have discovered that open standards can be more valuable than proprietary technology—that the company defining how the industry works is often more valuable than the company trying to own it.
For enterprise technology leaders evaluating AI investments, the message is straightforward: skills are becoming infrastructure. The skills organizations add to the skills today will determine how effectively their AI assistants perform tomorrow, regardless of the model that powers them.
The competitive battles between Anthropic, Openai, and Google will continue. But to the question of how to make AI assistants reliably good at specific tasks, the industry has quietly debated an answer — and it comes from the company that gave it.