Anthropic Launches Open Standard for AI 'Agent Skills' to Compete with OpenAI
# Anthropic Unveils Open Standard for ‘Agent Skills,’ Taking Aim at OpenAI in the Workplace AI Sector
**Tech firm releases new tools and an open standard as it seeks to enhance AI capabilities in business environments**
On Wednesday, Anthropic announced it would launch its "Agent Skills" technology as an open standard, positioning itself strategically against fierce competition in the rapidly evolving enterprise software landscape. This San Francisco-based artificial intelligence company also introduced comprehensive management tools for businesses and a directory of partner-developed skills from renowned firms such as Atlassian, Figma, Canva, Stripe, Notion, and Zapier.
This initiative marks a significant evolution of Anthropic's technology, which was initially revealed in October. What began as a niche development feature now has the potential to become a widely accepted industry standard.
“We’re launching Agent Skills as an independent open standard with a specification and reference SDK available at https://agentskills.io,” stated Mahesh Murag, a product manager at Anthropic, in an interview with VentureBeat. He highlighted that Microsoft, alongside several popular coding agents like Cursor, Goose, and Amp, has already adopted Agent Skills.
At its core, the skills function as organized folders containing instructions, scripts, and resources that direct AI systems on how to perform specialized tasks consistently. By packaging this procedural knowledge into reusable modules, users no longer need to generate complex prompts for every unique task they wish an AI assistant to carry out.
This innovation tackles a key weakness of large language models: while they boast extensive general knowledge, they often lack the specialized procedural expertise necessary for professional tasks. For example, a skill designed for PowerPoint presentations might include preferred formatting conventions and slide structures that the AI can access only when creating presentations.
Anthropic's system employs what it describes as "progressive disclosure," summarizing each skill efficiently in the AI's context window. Detailed information is retrieved only as required, allowing businesses to utilize extensive skill libraries without overwhelming the AI's working memory.
Already, Fortune 500 companies are incorporating these skills in various sectors including legal, finance, and accounting. The newly introduced enterprise management features enable supervisors on Anthropic's Team and Enterprise plans to centralize skill provisioning, ensuring that specific workflows are accessible while allowing employees to tailor their individual experiences.
“Enterprise customers are utilizing skills in both coding workflows and business functions like legal, finance, accounting, and data science,” Murag commented. He added that feedback has been overwhelmingly positive, as these skills help users personalize Claude to align with their actual work patterns, enabling quicker, high-quality outcomes.
The community's reaction has surpassed expectations, as Murag noted: "Our skills repository already crossed 20k stars on GitHub, with tens of thousands of community-created and shared skills."
Anthropic has launched with contributions from ten initial partners, creating a directory that includes prominent players in enterprise software. The collaboration with companies such as Atlassian, known for products Jira and Confluence, alongside design-oriented tools like Figma and payment services like Stripe, illustrates Anthropic's intention to connect Claude with existing business applications.
The partnerships are oriented more towards building an ecosystem rather than immediate revenue generation. “Partners who build skills for the directory do so to enhance how Claude interacts with their platforms,” Murag elaborated. "It’s a mutually beneficial ecosystem relationship similar to MCP connector partnerships. There are no revenue-sharing arrangements at this time."
In vetting new collaborators, Anthropic is adopting a cautious approach. "We began with established partners and are developing more formal criteria as we expand," he noted. "Our goal is to establish a valuable supply of skills for enterprises while spotlighting partner products."
Interestingly, the company is not imposing additional costs for these capabilities. "Skills work across all Claude surfaces: Claude.ai, Claude Code, the Claude Agent SDK, and the API. They are included in Max, Pro, Team, and Enterprise plans at no extra charge," Murag explained.
Their decision to release Skills as an open standard is a strategic calculation. By enabling skills to be transferable across different AI platforms, Anthropic believes that fostering ecosystem growth will ultimately be more beneficial than locking users into proprietary systems.
The strategy appears to be gaining traction. OpenAI has subtly adopted a similar architecture, as uncovered by developer Elias Judin, who found directories with skill files matching Anthropic's specifications in both ChatGPT and its Codex CLI tool.
This alignment suggests that the industry is converging on a shared solution for enhancing AI assistants’ abilities in specialized tasks without the need for costly model adjustments. The timing also aligns with broader standardization efforts within the AI sector, where Anthropic contributed its Model Context Protocol to the Linux Foundation last December, together with OpenAI, and co-founded the Agentic AI Foundation alongside Block.
Skills serve as a philosophical shift in the AI industry's approach. While traditionally there were specialized agents for different tasks, the Skills concept introduces a model wherein one general-purpose agent is empowered with a repository of specialized capabilities.
“This insight has significant implications for enterprise software development,” said Barry Zhang, an Anthropic researcher. “Instead of creating and maintaining various specialized AI systems, organizations can focus on developing and curating skills that encapsulate their institutional knowledge and best practices.”
Furthermore, a study conducted by Anthropic earlier this month highlighted that its engineers leveraged Claude for 60% of their work, yielding a reported productivity boost of 50%. Notably, a significant portion of Claude-assisted efforts contributed to innovative tasks that would have otherwise been deprioritized.
Despite these advancements, potential complications arise with the Skills framework. As AI systems gain capabilities, concerns about maintaining human expertise become prominent. Some engineers expressed worries about skill atrophy, emphasizing that while generating outputs becomes more accessible, learning remains essential.
Security risks are also a consideration. Skills grant Claude new functionalities through instructions and code, which could pose threats if malicious skills are introduced. Anthropic recommends using skills exclusively from trusted sources and advises a thorough audit of those from less familiar origins.
Lastly, with the open standard model comes governance challenges. Although Anthropic has detailed the specifications and released a reference SDK, the long-term management of this standard is yet to be determined. Whether it will be integrated into the Agentic AI Foundation or require its own governance framework remains an open question.
The evolution of Skills signals Anthropic's broader ambitions. Having transformed a developer-oriented feature into an essential standard embraced by major players like Microsoft and OpenAI, the company demonstrates that open standards can offer greater value than proprietary technologies.
For enterprise leaders assessing AI opportunities, the takeaway is clear: Skills are establishing themselves as foundational infrastructure. The expertise encoded into these skills today will dictate the effectiveness of AI assistants tomorrow, irrespective of the underlying model.
The competition among Anthropic, OpenAI, and Google will undoubtedly continue, but in addressing the challenge of ensuring AI assistants excel in specialized tasks, the industry has quietly unified around a solution pioneered by Anthropic.