Anthropic said Wednesday it will release its Agent Skills technology as an open standard, a strategic bet that sharing its vision for making AI assistants more capable will strengthen the company’s position in the fast-growing enterprise software market.
The San Francisco-based artificial intelligence company also unveiled organization-wide management tools for enterprise customers and a directory of partner-built skills from companies including Atlassian, Figma, Canva, Stripe, Notion, and Zapier.
The move represents a significant expansion of the Anthropic technology first introduced in October, turning what started as a niche developer feature into infrastructure that is now poised to become an industry standard.
"We are launching Agent Skills as an independent open standard with a specification and reference SDK available at https://agentskills.io." Mahesh Murag, product manager at Anthropic, said in an interview with VentureBeat. "Microsoft has already adopted agent skills within VS Code and GitHub; Hence there are popular coding agents like Cursor, Goose, AMP, OpenCode and others. We are in active interaction with others throughout the ecosystem."
Inside the technology that teaches AI assistants to perform specific tasks
Skills, at their core, are folders containing instructions, scripts, and resources that tell AI systems how to perform specific tasks consistently. Instead of requiring users to prepare detailed prompts every time they need an AI assistant to complete a particular task, skills package that procedural knowledge into reusable modules.
This concept addresses a fundamental limitation of large language models: while they have broad general knowledge, they often lack the specific procedural expertise needed for a specific professional task. For example, skills at creating PowerPoint presentations might include preferred formatting conventions, slide structure guidelines, and quality standards – information AI only loads when working on presentations.
Anthropic designed the system to live up to its name "Progressive disclosure." Each skill takes up only a few dozen tokens when summarized in the AI’s context window, with full details loaded only when the task requires them. This architectural choice allows organizations to deploy broader skill libraries without impacting the AI’s working memory.
Fortune 500 companies are already using skills in legal, finance and accounting
New enterprise management features allow Anthropic’s team and administrators of enterprise plans to centrally deploy skills, controlling which workflows are available across their organizations, while letting individual employees customize their experience.
"Enterprise customers are using the skills in production, both in coding workflows and across business functions such as legal, finance, accounting and data science." Murag said. "The response has been positive as the skills let them personalize the cloud to how they really work and get high quality output faster."
According to Murag, the community response has exceeded expectations: "Our skills repository has already crossed 20 thousand stars on GitHub, containing thousands of community-created and shared skills."
Atlassian, Figma, Stripe and Zapier join Anthropic’s skills directory at launch
Anthropic is launching with the skills of ten partners, a roster that sounds like experts in modern enterprise software. The presence of Atlassian, which makes Jira and Confluence, along with design tools Figma and Canva, payments infrastructure company Stripe and automation platform Zapier suggests that Anthropic is establishing skills as the connective tissue between the cloud and the applications businesses already use.
Business arrangements with these partners focus on ecosystem development rather than immediate revenue generation.
"Partners who create skills for the directory do so to enhance how the cloud works with its platform. This is a mutually beneficial ecosystem relationship similar to the MCP Connector Partnership," Murag explained. "There is no revenue-sharing arrangement at this time."
To vet new partners, Anthropic is taking a measured approach. "We started with established partners and as we expand, we’re developing more formal criteria." Murag said. "We want to create a valuable supply of skills for enterprises while helping partner products shine."
Notably, Anthropic is not charging extra for the capacity. "Skills work across all layers of the cloud: Cloud.AI, Cloud Code, Cloud Agent SDKs, and APIs. They are included in the Max, Pro, Team, and Enterprise plans at no extra cost. API usage follows standard API pricing," Murag said.
Why is Anthropic giving up its competitive advantage to OpenAI and Google?
The decision to release the skill as an open standard is a calculated strategic choice. By making skills portable across AI platforms, Anthropic is betting that ecosystem growth will benefit the company more than proprietary lock-in.
The strategy seems to be working. OpenAI has quietly adopted a structurally similar architecture in both ChatGPT and its Codex CLI tool. Developer Elias Judin explored the implementation earlier this month, finding directories containing skill files that mirror Anthropic’s specification – same file naming conventions, same metadata format, same directory organization.
This convergence shows that the industry has found a common answer to a complex question: How can you make AI assistants consistently good at a particular task without expensive model fine-tuning?
The timing is in line with broader standardization efforts in the AI industry. Anthropic donated its Model Context Protocol to the Linux Foundation on December 9, and both Anthropic and OpenAI co-founded the Agentic AI Foundation with Block. Google, Microsoft and Amazon Web Services joined as members. The Foundation will manage multiple open specifications, and the skills will fit naturally into this standardization effort.
"We have also seen how complementary skills and MCP servers are," Murag noted. "MCP provides secure connectivity to external software and data, while skills provide the procedural knowledge to use those tools effectively. Partners who have invested in strong MCP integration were a natural starting point."
The AI industry abandons specialized agents in favor of an assistant that learns everything
The skills approach is a philosophical shift in how the AI industry thinks about making AI assistants more capable. The traditional approach involved building specialized agents for different use cases – a customer service agent, a coding agent, a research agent. The skills suggest a different model: a general-purpose agent equipped with a library of special abilities.
"We used to think that agents would look very different in different domains," Anthropic researcher Barry Zhang said at an industry conference last month, Business Insider reports. "The agent below is actually more universal than we thought."
This insight has significant implications for enterprise software development. Instead of building and maintaining multiple specialized AI systems, organizations can invest in creating and curating skills that encapsulate their institutional knowledge and best practices.
Anthropic’s own internal research supports this view. A study published by the company in early December found that its engineers used the cloud in 60% of their work, leading to a 50% increase in self-reported productivity – a two- to threefold increase over the previous year. Specifically, 27% of cloud-assisted tasks involved tasks that would not otherwise be performed, including building internal tools, creating documents, and addressing tasks called in by employees. "paper clippings" – Small improvements in the quality of life that were consistently denied priority.
Security risks and skills erosion emerging as concerns for enterprise AI deployment
The skills framework is not without potential complexities. As AI systems become more capable through skill, questions arise about maintaining human expertise. Anthropic’s internal research found that while skills enabled engineers to work across more domains – backend developers were building user interfaces, researchers were creating data visualizations – some employees were concerned about skills erosion.
"When it’s so easy and fast to produce output, it becomes harder and harder to find time to actually learn something," said an Anthropic engineer in an internal survey of the company.
There are also security considerations. Skills provide new capabilities to the cloud through instructions and code, meaning malicious skills could theoretically introduce vulnerabilities. Anthropic recommends installing skills only from trusted sources and performing a thorough audit from less-reliable sources.
The open standards approach also presents governance questions. While Anthropic has published the specification and launched a reference SDK, long-term management of the standard is undefined. Whether this would fall under the Agentic AI Foundation or require its own governance structure is an open question.
Anthropic’s real product may not be the cloud – it may be the infrastructure on which everything else builds
The skill’s trajectory reveals something important about Anthropic’s ambitions. Two months ago the company introduced a feature that looked like developer tools. Today, that feature has become a specification that Microsoft builds into VS Code, which OpenAI replicates in ChatGPT, and which enterprise software giants rush to support.
This pattern echoes strategies that have previously shaped the technology industry. Companies from Red Hat to Google have found that open standards can be more valuable than proprietary technology — that a company defining how an industry works often gets more value than a company trying to make it entirely its own.
For enterprise technology leaders evaluating AI investments, the message is simple: Skills are becoming infrastructure. The expertise that organizations encode skills today will determine how effectively their AI assistants will perform tomorrow, regardless of which model powers them.
The competitive battle between Anthropic, OpenAI and Google will continue. But on the question of how to make AI assistants reliably good at specific tasks, the industry has quietly agreed on one answer — and it comes from the company that gave it.
<a href