AI Skills Offer Personalized Assistance, Streamlining Tasks
Translated & summarized from Vesty by baba
Artificial intelligence "Skills" allow users to create personalized AI assistants by providing sets of instructions and knowledge, streamlining repetitive tasks. Developed by companies like Anthropic, OpenAI, and Google, these skills can be process-oriented or information-oriented. Software engineer Alex Khoroshin has created a library of over 200 skills for Israeli users. While skills can be transferred between platforms, users must be cautious of potential security risks and grant minimal permissions.
The story in 6 lines · by baba
- AI "Skills" enable personalized assistants by providing instructions and knowledge, reducing repetitive task explanations.
- Anthropic, OpenAI, and Google have introduced skills, which can be process-oriented or information-oriented.
- Alex Khoroshin maintains Skills IL, a library of over 200 skills for Israeli users.
- Skills offer advantages over prompts, allowing for complex workflows and portability across AI tools.
- Users can create skills directly within chatbots or leverage advanced features like connectors and automation.
- Security is a concern, as skills can execute code; users should review instructions and permissions carefully.
Artificial intelligence tools like ChatGPT, Claude, and Gemini often require users to repeatedly explain the same tasks, such as drafting emails or summarizing meetings. A new feature called "Skills" aims to solve this by allowing users to create personalized AI assistants with specific instructions, knowledge, and tools. This "knowledge pill" enables the AI to become an expert in a particular area, eliminating the need to start from scratch with each interaction.
Anthropic, the developer of Claude, was among the first to introduce skills in October 2025, followed by OpenAI and Google. These skills can be "process-oriented," guiding the AI through a sequence of actions, or "information-oriented," providing specialized knowledge. While the distinction isn't always strict, process skills are becoming more relevant as AI models gain more inherent knowledge.
Skills offer significant advantages over standard prompts. They transform repetitive tasks into persistent instructions that the AI can activate with a simple trigger, and they enable the creation of complex workflows, even chaining multiple skills together. Unlike proprietary "projects," the open SKILL.md format allows for greater portability across different AI tools, though some adjustments may be necessary.
Creating a skill is often possible directly within the chatbot interface. Users can perform a task, then ask the AI to format the process as a skill, receiving a downloadable ZIP archive. Alex Khoroshin, a software engineer and owner of Yootech, has developed and maintains Skills IL, a library of over 200 skills tailored for Israeli users, covering tasks from insurance comparisons to programming tools. He emphasizes clear descriptions, avoiding unnecessary instructions, and providing examples for effective skill creation.
Advanced users can create meta-skills that coordinate other skills or combine them into plugins for complex workflows. AI tools can also be granted access to external services like Gmail or Google Sheets via connectors, allowing skills to gather data, generate reports, and send emails automatically. Automation is also possible, with tasks scheduled for regular execution.
While skills can be transferred between platforms like Claude, ChatGPT, and Gemini, compatibility issues may arise due to differing feature support. Users are cautioned about security risks, as skills can potentially execute harmful code. Khoroshin advises scrutinizing skills from trusted sources, reviewing their instructions and requested permissions, and granting only the minimum necessary access.