Markdown Converter
Agent skill for markdown-converter
Beyond text, image, audio, or video generation from direct user prompts, AI agents based on generative AI
Sign in to like and favorite skills
Beyond text, image, audio, or video generation from direct user prompts, AI agents based on generative AI pose both challenges and opportunities.
Broadly defined, AI agents are software systems or programs that use AI models to reason and initiate actions and tasks toward one or more goals. In more complex workflows, multiple AI agents coordinate work towards shared goals for the system.
Web browsers that embed agents, like Claude's Chrome browser extension1, allow these agents full access to the saved passwords and websites and can initiate actions like filling out forms and performing other interactions with websites. Already these AI browsers present significant concerns in higher education where these browsers can log into learning management systems, respond to discussions, take quizzes, and simulate other student behavior with little or no direction by the student.2 These new AI browsers present significant security threats and potential legal liability but also offer significant time savings and productivity boosts for users automating their personal and professional workflows.
In the past year, AI bots have been overwhelming library and other cultural heritage websites3, harvesting material including catalogs, institutional repositories, and other assets. Rosalyn Metz in her blog post, When Bots Meet Books4, she recommends that libraries provide clear access points for AI bots, restrict access to bad bots, encourage attribution of library collections, and even explore new revenue models with AI companies.
AI Researcher and pundit Andrej Karpahty in a recent presentation5 talks about how software development has entered a new phase where we need to think of LLMs as analogous to computer's operating systems and prompts (through AI Agents) are computer code. For libraries and other cultural heritage institutions to pivot services and collection for access and operations includes both internal and external AI Agents. One suggestion he recommends is to create specific markdown files that expose the content specifically for AI agents in an
llms.txt5 files,
similar to a sitemap or robots.txt that directs LLMs to what content that to be harvested and
ideally, reduce the traffic and demands on institutional web sites.